Ecological volatility and human evolution: A novel perspective on life history and reproductive strategy

Building on work by others,1-3 this perspective contrasts with a previous tendency to emphasize systematic climate trends as the key ecological stress driving hominin adaptation. Following the acquisition of detailed information on global climate over the last 5 million years (for example, ice-core isotope records and deep-sea sedimentation rates), much attention was directed to the hypothesis that hominin traits were shaped by trends toward a cooler and drier environment, during which the African savannah habitat expanded substantially. Vrba's4 “turnover pulse” hypothesis, for example, linked temperature shifts with the pattern of evolutionary extinctions and speciations, including those in hominins. Isotopic biomarkers of plant species also indicate the expansion of savannah grasses across East Africa.5 The emergence of the savannah niche was thus considered the primary ecological stress shaping hominin evolution, with long-term cooling and greater aridity treated as influential ecosystem stresses. High daytime temperatures were also considered to be important.4, 6, 7

A novel counter-argument is that climatic instability, rather than any specific environment, was the key challenge facing hominins.1, 2 Focusing on multiple rather than single indicators of paleoclimate to gain a more integrated perspective, Potts3 has argued that ancestral environments are best characterized as oscillating between states of high versus low volatility, and that there were numerous lengthy periods of high volatility within the last 3 million years. Potts proposed that many hominin adaptations were therefore ways of reducing commitment to any single environment, instead aiding adaptation to a range of environments or changing environments. Figure 1 illustrates ¹⁸O isotopic data over the last 5 million years, providing a proxy for mean annual temperature.8 While an overall trend to cooler temperature can be discerned, arguably more notable is the increasing variability of climate, reflected in arid-moist as well as temperature cycles.3

Such climate volatility imposes substantially greater stochasticity in local ecological conditions. The pattern of change is assumed to be unpredictable and complex, as different species respond at different rates.2 Paleoenvironmental evidence is now emerging of rapid and profound shifts in habitat quality in the eastern African regions where the evolution of Homo is thought to have occurred, as well as in other global regions relevant to understanding broader hominoid evolution, such as Asian forests.3, 9

Some of this climatic variability can be attributed to predictable geophysical phenomena, in particular irregularities in the earth's orbital path, known as Milankovitch cycles, which reflect variability in the exposure of the earth's surface to insolation. There are three main such cycles, which occur with periodicity of ∼100,000, ∼41,000, and ∼23,000 years, and are discernible in the isotopic record.10

While Milankovitch cycles ensure that the earth's climate is never static, ecological shocks contribute additional volatility. These include volcanic eruptions, meteorite strikes, and other forms of global weather instability.2 Recent analysis of paleoenvironmental data from Hadar, Ethiopia, suggests that Australopithecus afarensis was already able to tolerate directional cooling, climate stability, and high variability over 0.5 million years.11 The expanded range of the genus Homo suggests that this tolerance of variable environments became enhanced, but how did such tolerance evolve?

LIFE-HISTORY THEORY

Evolutionary economics is a theoretical perspective emphasizing the way selection imposes rationality on organisms' biology.12 Life-history theory applies this approach to life cycle strategies by assuming that energy is a scarce resource, so that investment of energy in one function reduces its investment in others.13 Maintenance, growth, immune function, and reproduction are considered the primary functions competing for energy supply.13

By convention, life-history theory assumes that energy supply and mortality risk are the two principle stresses, shaping the pace of maturation and reproductive strategy.14 Thus, under conditions of high mortality risk, organisms are predicted to grow fast and to mature early. Any increase in the availability of resources is then likely to be directed to increasing the number of offspring produced. As the risk of mortality decreases, organisms can afford to grow for longer and become bigger. This allows them to specialize to specific environments, while also increasing the energetic efficiency of reproduction, which in turn increases the survival chance of each offspring. Larger, slow-growing, and specialized organisms therefore produce fewer offspring, but increase the amount of parental care per offspring.15 To some extent, these varying strategies shape the genome of each species and account for species-wide differences in life-history profile.16

Within these general strategies, each species also demonstrates variable degrees of phenotypic plasticity. Through norms of reaction,16 organisms can develop different phenotypes according to environmental conditions. Thus, within any given species, growth may be faster if energy supply is increased.14 Likewise, if mortality cues are greater than typical, an organism may mature faster to breed sooner.14 The capacity for such plasticity varies substantially across species, a component of variability itself assumed to have been shaped by selection.

Such sensitivity to cues of energy availability and mortality risk is generally assumed to allow information to enter the phenotype during the life-course, especially during “critical windows” in early life, when plasticity is greatest.17 Mechanistic studies show that developmental plasticity is enabled by several physiological systems, including growth variability, the programming of hormonal axes, effects on brain structure, and epigenetic alterations in DNA expression.18 Many of these early physiological adaptations track into later life19 and may affect subsequent generations.

However, some ecological factors exert a profound influence on survival and reproductive fitness and yet occur irregularly.20 Ecological shocks may occur frequently and yet pass some generations by. This poses several specific evolutionary problems. How should such events be accommodated when they do occur? And how can any adaptive responses be incorporated into the genome when not all generations are exposed?

THEORETICAL APPROACHES TO ADAPTATION TO EXTREME EVENTS

According to Vermeij,12 extreme events induce two kinds of adaptive biological responses within organisms' life spans. The first type, passive resistance, involves closing down all nonvital functions until improved environmental conditions allow reactivation.12 The second type, active responses, includes a variety of ways whereby organisms attempt to avoid or modify their environment.12 Uncertainty regarding future conditions also favors bet-hedging and phenotypic flexibility.14, 21

Crucially, however, tolerating ecological adversity is only one of two key ways organisms respond to extreme events. Because climate in each region is inconsistent, times of adversity tend to alternate with times of profligacy. For example, many contemporary species that experience harsh regular droughts from ENSO events target their reproduction at intervening periods of high rainfall, when the supply of food spikes.20 Rapid breeding during periods of high resource availability is therefore just as important as the ability to withstand harsh environments.

Moreover, faster reproduction offers a solution to the second dilemma, how to develop adaptive strategies when not all generations experience the stress. Reproductive flexibility distributes the adaptive response across rather than within generations, thus making a lineage more robust in response to adverse experiences within any single generation. Being able to upregulate or downregulate reproduction according to the availability of resources is therefore a more flexible adaptive strategy than evolving a specific physical or physiological trait.

EL NINO/SOUTHERN OSCILLATION (ENSO) EVENTS AS A MODEL OF CLIMATE SHOCKS

In the last few thousand years, many global regions have experienced frequent climate disruptions, which manifest as extreme drought or rainfall linked by oceanic temperature oscillations.22 The current form of ENSO is thought to have persisted for 5,000 years, but longer records indicate catastrophic floods for 40,000 years in Peru and effects on fossil coral in Sulawesi for 125,000 years, from immediately after the last interglacial.20 In other global regions with especially powerful ENSO impacts, evidence of biota adaptations suggests that climate disruptions have an extensive history. Indeed, some form of ENSO appears to date back ∼5 million years.20, 23

Many species of pelagic seabird are particularly vulnerable to the fluctuations in fish stocks induced by ENSO cycles.24 The availability of nutrients in oceans is strongly associated with temperature. The highest density of nutrients is generally found in cooler, deeper waters, where the majority of dead organisms decay and release nutrients back into the chain of supply.12 However, the world is characterized by five locations of particular productivity (off Peru, California, Mauritania, Namibia, and Somalia), where upwellings force this nutrient-rich water to the surface. In terms of fish yield, these five areas are approximately 65,000 times more productive than is the open ocean, despite comprising only 0.1% of the total ocean surface area.20 These areas allow fish to thrive on the nutrients, in turn stimulating massive populations of seabirds whose demography then tracks that of the fish.

So long as upwellings persist, the seabird populations boom, but during ENSO events they can crash spectacularly, with mass mortality of both parents and offspring. Several features of seabird life history are now considered to have evolved in response to this specific stress. The longevity of these species, stretching to 3 to 5 decades, allows adults to miss some breeding seasons (“reproductive sabbaticals”) but still produce offspring over the total life course.24 Several factors increase the likelihood of success in poor conditions: clutch size is small, typically 1-2 eggs, and the chicks typically have a slow rate of growth, often reaching reproductive maturity only at around 5 years.24 Growth is further slowed through life-course plasticity when fish stocks are low. During good conditions, however, chick survival is very high, with fledging success rates often around 85%-90%.24 Breeders tend to have higher blood lipid levels, demonstrating the value of energy stores in reducing the risk of missing a breeding season.25 This life-history strategy allows the clustering of parental investment within brief periods of good conditions relative to the long life spans that enable the toleration of poor conditions.

In line with theoretical predictions, a wide range of organisms show diverse methods for “sitting out” the droughts imposed by ENSO events. Strategies for evasion include birds migrating, invertebrates producing larvae that remain dormant until precipitation returns, amphibians burrowing deep into the soil and sealing their skin, iguanas shrinking in body size, and ephemeral plants setting seeds and then dying so that the lineage is preserved in the dormant seeds.20 Other species show various strategies for tolerating the atypical conditions, including alterations in basal metabolic rate, diet, and foraging strategy.20 An increased capacity for migration clearly reduces the risk of extended exposure to environmental disruptions. In Australia, characterized by regular ENSO events, over one-quarter of bird species are nomadic and hence can opportunistically track the best resources.20

When conditions improve, many species show an extraordinary capacity to respond rapidly and opportunistically by stimulating reproduction. Studies of birds regularly exposed to drought illustrate substantial plasticity in reproductive effort, through flexibility with respect to when clutches are laid and clutch size. Adult phenotype also often shows a capacity for semi-dormancy during harsh conditions. In Galapagos finches, gonad mass contracts during dry spells and increases again with the onset of the rains, allowing the birds opportunistically to tailor reproduction to the period in which food for their offspring is available.26 Zebra finches in Australia likewise maintain their ovaries in a semi-developed state, but reproductive maturation can occur within 2 weeks of the arrival of favorable rainfall conditions.27 These birds have been observed to initiate courting within hours of the return of the rains, and in good conditions can breed without stopping for 11 months.28

Cooperative breeding is another means of maximizing reproductive output in good conditions. Approximately 12% of continuously breeding Australian species breed cooperatively, a proportion five times the global average. High levels of cooperative breeding have also been observed in southern Africa, another region of environmental unpredictability.29 A recent analysis of birds found a strong global correlation between ecological stochasticity and the prevalence of cooperative breeding.30

An alternative “conveyor belt” strategy for upregulating reproduction in good conditions has been observed in kangaroos.20 Although each offspring takes ∼600 days to mature, kangaroos can invest in three young at a time, producing a new offspring every ∼240 days. The initial maternal energy investment in the embryo, which weighs ∼1 g at birth, is minimal. Accordingly, the conveyor belt can be rapidly halted with little waste of investment.20 Marsupials in general also have lower basal metabolic rates than do other mammals of comparable weight, occupying predictable seasonal environments in higher latitudes.31 This trait is thought to explain their persistence in ENSO-affected Australia, where placental mammals failed to remain viable.20 However, marsupials are able to raise their metabolic output above that of nonmarsupials during reproduction, and hence provide a rapid burst of reproductive investment.31 Like the bird adaptations described earlier, this “stacking” system enables rapid opportunistic responses to sudden improvements in ecological conditions.

These characteristics are strikingly similar to many of the Homo life-history traits listed at the start of this article. Building on previous discussion by Potts,1, 2 who emphasized general ecological instability as a key hominin stress, I suggest that the traits I will discuss may be considered potential life-history adaptations to irregular but frequent ecological shocks similar to ENSO cycles. The particular emphasis here is on the capacity to tolerate periods of poor conditions and the capacity to cluster high reproductive output during periods of good conditions.

HYPOTHESES FOR THE EMERGENCE OF HOMO LIFE HISTORY

Relatively long life spans and late maturity are characteristics of all extant ape species, indicating that a slow life history was the ancestral state of this group of mammals.32 However, variation in the timing of brain and dental development does not account for between-species variability in life-history profiles, and most human life-history traits represent extreme values. Several hypotheses proposed to account for this unusual Homo profile are briefly reviewed here.

A lengthy life span has generally been attributed to a reduction in mortality, on the basis that the rate of senescence decreases in correlation with adult mortality risk.33 This, in turn, is explained by the “disposable soma” theory, which holds that reproduction and cellular maintenance compete for energy and that earlier breeding accelerates the rate of aging.34 Thus, a reduction in mortality risk in humans relative to chimpanzees has been proposed to have been central to greater longevity in Homo.35 However, decreased mortality risk is not the only factor favoring a slower rate of aging. A second potential factor is an increase in fecundity with increasing age.33 Such an effect is predicted even in determinate-growth mammal species that do not change in size after maturing if, through changes in reproductive capacity, they produce more offspring per unit of time later rather than early in adult life. This insight was invoked by Hawkes in her model of the evolution of postmenopausal longevity, or grandmothers, in humans.15

Consistent with Hawkes' theoretical approach, much emphasis has recently been placed on cooperative breeding as a core Homo life-history adaptation. Hrdy, for example, proposed that cooperative breeding increased reproductive fitness by allowing reduced birth intervals, and hence stacking, without compromising the rate of infant survival,36 thus making a connection between several characteristic Homo traits. Grandmothers, and hence longevity, also fit with this model as key providers of assistance to mothers.37 Importantly, Hrdy suggested that cooperative breeding could have preceded encephalization.36 However, although she reviewed compelling evidence that humans are cooperative breeders, Hrdy referred only briefly to broader ecological stresses favoring this reproductive strategy,36 leaving it unclear how grandmothers evolved.

Others have focused on the slow growth of humans compared to chimps following weaning.38 This growth pattern has, by convention, been linked to the large brain, though for different reasons. Some have argued that growth is delayed to increase learning time. For example, Kaplan, Lancaster, and Robson39 have argued that hunting is a complex behavior and that an extensive period of learning is required to achieve proficiency. Others, focusing on the energy costs imposed by the large brain, which are compounded when mothers simultaneously provision several offspring, have proposed that slow growth reduces these costs to protect brain growth while also constraining the total maternal energetic burden.40

The notion that cycles of feast and famine were an important stress in human evolution was encapsulated in Neel's thrifty genotype hypothesis, which suggested that human populations evolved genetic variability in metabolism through different degrees of exposure to famine.41 This hypothesis has been criticized, not least because famines have affected all human populations in recent millennia.42, 43 A more plausible hypothesis is that oscillations in energy availability were a major ecological stress for the genus Homo, shaping a variety of components of phenotype.44

Neither the cooperative breeding hypothesis nor the slowed growth hypothesis clearly identifies external ecological stresses that might have favored such adaptations in humans in contrast to chimpanzees and gorillas. Indeed, many authors see the brain as a key internal source of stress on energy allocation trade-offs.39, 40 I will review ENSO as a plausible external stress favoring stacking and cooperative breeding, as well as related physiological traits, including slowed maturation and adiposity.

THE ENSO HYPOTHESIS FOR APE LIFE-HISTORY DIVERSITY

It is difficult to reconstruct ancestral ape environments, but I assume that chimpanzees and gorillas were less exposed to ENSO due to their reduced occupation of more open environments with longer average dry seasons, which are believed to have shaped the evolution of Homo.45 Chimpanzees occupy a range of ecotypes, ranging from rainforest to woodland savannah, and demonstrate variable behavior across these niches.46 Nevertheless, the range of ancestral and contemporary chimpanzees appears to have been narrower than that of hominins due to their reduced ability to tolerate more arid environments,45 where ENSO effects are especially strong. Stochasticity in the availability of fruit, a key food source for all apes, is much greater in the Asian forests occupied by orang-utans than it is in the African habitats occupied by chimpanzees and gorillas.9, 47 Using chimpanzees and gorillas as a “basal ape” reference point, I assume that both orang-utans in Asia and Homo in Africa have experienced greater exposure to ENSO.

Based on the theoretical perspective and empirical evidence I have discussed, I review the following predicted adaptations to ENSO exposure across apes: facultative conception and acceleration of reproduction in good conditions; cessation of reproduction and physiological resilience in tough conditions; and slowed growth and lengthening of the life span.

My primary prediction is therefore that high exposure to ENSO in African environments was a major factor shaping the characteristic life-history profile of Homo.

REPRODUCTION: FACULTATIVE RESPONSES AND THE ABILITY TO BOOM

I assume the “basal ape” to be similar to contemporary chimpanzees in terms of body size, brain size, and interbirth interval (∼5 years). The two theoretical predictions for ENSO exposure are increasing reproductive effort in good condition, and reducing reproductive effort in poor conditions.12

Most apes appear to have a facultative reproductive response.48 Chimpanzees breed faster in captivity, indicating energetic constraints in the wild, whereas gorillas appear to be less affected by seasonality in food supply and may reproduce at near-optimal rates year-round.47 Recent work has demonstrated ecologically induced variability in estrogen concentrations in chimpanzees, orang-utans, and humans.48-50 However, while this common pattern suggests that the “basal ape” can respond to ecological signals in terms of conception, the situation is very different regarding the total rate of reproduction. Table 1, which illustrates key life-history traits for apes and humans, indicates that the interbirth interval in humans is the shortest, below half that in orang-utans. ENSO exposure, in interaction with other socio-ecological parameters, may explain these contrasts.

The life-course period when ape mortality is greatest tends to be early life although, as Table 1 shows, that is variably so between species. Ecological stochasticity exacerbates early-life mortality by making it harder to meet offspring energy requirements. This may be addressed in two ways: buffering the stochasticity to constrain offspring mortality or increasing the production of offspring to counteract offspring mortality.14 Which of these strategies is adopted depends on whether resolving the uncertainty in energy supply is adequate to buffer ENSO or whether external mortality risk is also raised. Intriguingly, the marked differences in interbirth interval between humans and orang-utans suggest very different adaptations.

The long interbirth interval in orang-utans is consistent with the low infant mortality rate, suggesting that the key stress is energetic, favoring slowed offspring growth along with other maternal adaptations. Some authors speculate that ancestral Pongo may initially have lived in small gorilla-type groups, but that increasing uncertainty in energy supply from ENSO obliged a more solitary reproductive strategy.9 Lacking the possibility for alloparental care, orang-utans therefore appear to have increased the interbirth interval to an extreme degree in order to reduce the energetic burden of the offspring.

In contrast, humans reproduce faster than the “basal ape,” a strategy that may be the driving force behind other species-contrasting adaptations. Faster reproduction, on its own, would impose unviable energy costs on the mother. These costs appear to have been met in several related ways, as I will review: stacking of offspring so that rapid reproductive output does not come at the expense of offspring care, cooperative breeding (that is, a social redistribution of the total costs of reproduction), slower offspring growth, and increased adult body size and energy stores (Box 2). While the beneficial effects of short interbirth intervals for total fertility rate seem obvious, little attention has been directed to potential costs: any population continuously experiencing positive growth rates must rapidly explode and become an unsustainable stress on its environment.15 To counter persistent population growth, the capacity for high fertility must have been balanced for the majority of human evolution. It has generally been assumed that the balancing stress was mortality, particularly in infancy, as to some extent supported by the data in Table 1. However, contemporary human foragers have infant mortality rates similar to those among chimpanzees and gorillas, suggesting that mortality differences are not the only explanation for reduced interbirth interval in humans. The other factor that can balance high fertility capacity is an increased risk of conception failure due to energy scarcity during some periods.

Box 2. Filling in the Gaps: A “Cooperative Conveyor Belt” Reproductive Strategy

How can more offspring be produced? One way would be to increase clutch size; that is, to increase the likelihood of twins. However, twins generally have poor survival, and both might die in a single extreme event. Although twinning does occur at a low rate in humans, and increases maternal fitness,94 increased clutch size is not a viable primary strategy for increasing human reproductive rate.

Figure 4 shows that as juvenile mortality increases above a typical ape rate of 40%, either the duration of the female reproductive career must increase or mean interbirth interval must decrease to allow the average female to produce two surviving offspring. The duration of the human reproductive career is similar in humans and chimpanzees. Reducing the mean interbirth interval, in combination with slowed growth, stacking, and cooperative care, seems the human solution. Several offspring could be conceived and gestated within a “good” decade even though each offspring takes ∼two decades to reach maturity.

4. Simple simulation illustrating the effect of increasing offspring mortality rate on interbirth interval and duration of reproductive career in order to produce two surviving offspring. The model assumes that offspring are independent at the end of the interbirth interval and that offspring mortality occurs on average halfway through the typical interbirth interval, at 2 years postconception. A standard ape mortality rate of 40% would be consistent in human forager populations with a mean interbirth interval of ∼4 years and a mean reproductive career of ∼12 years. As offspring mortality rises, the interbirth interval must decline or the reproductive career must increase. Stacking represents a solution for accommodating a greater number of significant increases in offspring mortality without requiring substantial increase in duration of the reproductive career.

Opportunistic breeding is favored by the powerful influence of energy stores and associated hormones on the mechanisms regulating conception.48 During the 1974-1975 famine in Bangladesh, for example, fertility declined by 34%, but then increased above the long-term background rate by 17% in the postfamine period.95

This composite strategy of reproductive rate tracking ecological conditions and producing offspring relatively rapidly in periods of good conditions shows similarity with other ENSO-exposed species. Though on a slower time-scale, humans have a conveyor-belt strategy, as do kangaroos.20 Humans maximize nutritional investment in offspring in early life when mortality risk is greatest, while distributing juvenile growth costs over an extended total period and meeting these costs through cooperative care.

Together, reduced interbirth intervals and stacking may therefore represent a key solution to ecological volatility, enabling the rapid production of offspring during bountiful conditions to counter periods when conception is delayed. With cooperative breeding unviable, orang-utans adopted a contrasting reproductive strategy of producing offspring very slowly, which appears to be successful as a result of low levels of infant morality.

COOPERATIVE BREEDING

Hrdy has made a strong case for humans being cooperative breeders.36 Forager societies show high levels of prosocial behavior in relation to offspring care, with a variety of allomothers providing the mother with support of both food resources and “activity subsidies” (for example, babysitting). In contemporary small-scale societies, juvenile females and grandmothers contribute alloparental care, as do many men (though not necessarily fathers or grandfathers) and other maternal kin.37

The critical role of cooperative breeding in human reproduction has been demonstrated by simulations showing that mothers could not meet the energy requirements of their offspring without such support.⁹⁹ Even taking into account physiological adaptations in the offspring (slowed growth)40 and mother (larger body size, adiposity),44, 51 the demand imposed on the mother by stacked offspring amounts to an unviable individual energy burden. But what ecological stress faced by the “basal ape” favored the emergence of stacking and cooperative breeding? The most plausible driving factor was the reduction in interbirth interval, as I proposed earlier.

Through stacking, for much of the reproductive career mothers and their helpers may provision several offspring of different ages and needs at any given time. This resembles the “conveyor belt” breeding strategy of kangaroos, enabling several offspring to be launched within the time the first one takes to reach maturity. The slow growth rates of early offspring are therefore traded off against the mother's investment in further offspring.

The benefits of cooperative breeding are enhanced by diversity in foraging strategies, which allow flexibility in those contributing to offspring care while introducing further resilience against ecological stochasticity. The pursuit of different food resources by different members of a social group allows some individuals to be relatively unproductive at any given time-point, their shortfalls being met by other group members. Such bet-hedging across multiple food resources is suggested in the archeological record by the emergence of central food processing locations by 2 mya.3

Thus, in the face of uncertainty in energy supply, relatively solitary female orang-utans have “hunkered down” and produce single offspring very slowly, whereas humans, with the additional stress of higher infant mortality rates but less constraint on sociality, have become cooperative breeders. These reproductive strategies are associated with physiological adaptations in growth, body composition, and aging.

SLOWED GROWTH AND LONGEVITY

Compared to the “basal ape,” both humans and orang-utans show a slower pattern of maturation, which reduces the burden on the maternal energy budget. The initial selective pressure favoring slowed growth is likely to have been its role in reducing the maternal reproductive energy burden. However, as discussed earlier, slowed growth may also have reduced the intrinsic rate of aging and thereby enabled the evolution of greater longevity and reserve capacity.33, 34, 52 The greater longevity of humans compared to orang-utans52 could therefore arise from stronger selection on delayed maturation in humans, viable because of alloparental care of older offspring.

Recent research suggests that longevity increased systematically in early Homo relative to that of australopithecines,53 although the exact pattern of this increase remains uncertain because of the lack of direct fossil evidence.54, 55 Greater longevity may have benefited humans partly through modest increases in mean reproductive career, but it may also have increased the survival of later-born offspring through grandmothering, itself enabled by postreproductive longevity. This suggests that reduced interbirth interval, slowed growth, and stacking may have evolved first in Homo, presumably supported through cooperation among young adults and adolescents, with grandmothering possibly becoming important more recently as postmenopausal longevity evolved.

Growth is not only slow generally,56 but also highly plastic in early life. Under poor conditions, infants grow slowly, reach puberty later with lower adiposity, and have reduced adult size.57-59 Under bountiful conditions, infants grow rapidly and reach puberty earlier with greater adiposity.60, 61 During childhood, harsh conditions can temporarily halt growth, but when good conditions return the deficit can be recovered through rapid catch-up growth.62 Mortality rates in early life closely track growth variability, so that there is a well-established association between birth weight or infant weight and risk of mortality.63, 64 Offspring have better survival when they are larger, but such faster growth imposes a larger burden on those providing the energy.

These effects show intriguing similarity to those observed in seabirds, where early growth slows markedly under conditions of energy stress, whereas fledging rates are very high under good conditions.24 Unlike childhood growth, which is hormonally regulated, infant growth appears tailored to variability in early-life parental provisioning capacity.19 Parents with reduced energy-capital produce smaller offspring.65 In famines, infants, children, and the elderly comprise the majority of those who die,66 highlighting the greater resilience of breeding-age adults which have greater likelihood of sitting out the tough conditions. In good conditions, however, the reproductive “conveyor belt” can be rapidly accelerated through cooperative breeding, and plasticity allows individual offspring to mature faster. Like seabirds, human populations can grow rapidly when infant mortality falls, as shown by recovery from population crashes.67

ADAPTATIONS IN BODY COMPOSITION

Intuitively, stochastic environments might be assumed to favor small body size. However, because of the negative allometric association between absolute body mass and energy requirements per kg,68 larger animals can eat diets of poorer nutritional quality.9 Furthermore, large animals can carry larger absolute masses of adipose tissue, which can buffer the body's energy needs for longer periods.69

Both humans and orang-utans may draw on fat reserves to buffer temporary inadequacies in daily energy income. Fat can extend survival during starvation, especially for females.70 However, given that starvation leads to the death of both sexes within a few months, typically from infectious diseases,70 the fitness value of fat is likely to have been greatest in enabling toleration of lengthy periods of energy scarcity, while also enabling rapid accretion of energy reserves when food supplies increased.

Humans can rapidly store energy as fat in diverse circumstances. For example, when malnourished women with eating disorders regained weight, the overall 28% increase in weight comprised a 21% increase in muscle mass but 124% increase in total fat mass and a 174% increase in visceral adipose tissue,71 the adipose depot where immune system genes are most strongly expressed.72 A high rate of fat accretion may also occur during catch-up growth following early-life malnutrition.73 Furthermore, adequately nourished adults can also gain weight very rapidly. Over 2 months, voluntary participants in the Guru Walla ceremony in Cameroon undergo a 25% increase in weight, of which ∼70% comprises fat.74

This capacity to gain fat in bountiful times, as well as to oxidize it during periods of inadequate food intake, means that fat offers the capacity not only for resilience during ecological shocks, but for rapid reproduction during good conditions.48, 75 The availability of oxidizable fuels is an established determinant of fertility in mammals.76 Adiposity and leptin contribute to this regulatory mechanism, thus linking reproductive function closely with energy availability.48, 77 In predictable seasonal environments, primates will ideally match the timing of delivery to the timing of maximal fruit availability because lactation is the most expensive component of reproduction. However, with greater uncertainty in energy availability, apes are obliged to match conception to energy supply.47 Even among cooperative breeders, it is still individuals that must add each new offspring to the conveyor belt, which explains the sensitivity of conception to ecological conditions. This scenario, then, requires an alternative strategy for meeting the high costs of lactation. Again, adiposity plays a key role, as demonstrated in both orang-utans and humans, in which it enables continuing ecological perturbations to be smoothed over.50, 78, 79

Adiposity similarly promotes reproduction in ENSO-exposed seabirds.25 Furthermore, experimental studies on seabirds show that energy scarcity induces adults to preserve their own body mass at the expense of their chicks' survival, though in sex-specific ways.80 Thus, while energy is generously transferred to offspring in good conditions, it is withheld in poor conditions. The increased risk of stillbirth during famines81 and the sensitivity of offspring growth to maternal capital65 suggest that humans have a similar facultative mechanism.

In contrast with other apes, humans show one additional adaptation, high infant adiposity. Human infants are born fatter than those of most other species,82 although the magnitude of this effect remains uncertain. The neonatal percentage of fat averages ∼14% in industrialized populations,44 but was ∼8% in a large study of Ethiopian newborns.83 One hypothesis is that this neonatal adiposity co-evolved with larger neonatal brain size.82 However, it is also possible that neonatal adiposity promoted survival before the evolution of larger brains. For example, adiposity in chicks is also recorded in many species of seabird exposed to ENSO, where it helps buffer uncertainty in parental food supply.84 Intriguingly, birth weight appears to have been relatively large in australopithecines,85 despite ape levels of encephalization, raising the possibility that neonatal adiposity was already enhanced at this stage of hominin evolution in response to ecological stochasticity.

Once again, therefore, humans and orang-utans show both a similar strategy, using adiposity to buffer energy stochasticity, but also a difference, in that human neonates appear to acquire a privileged supply of energy for the most vulnerable period of the life course. Once present, this extra fat may have facilitated the emergence of infant encephalization under positive feedback mechanisms.

CONCLUSIONS

Many authors have assumed that human life history has been dominated by the expensive brain, so that the species is seemingly subject to unique selective pressures. Many authors have argued that lengthening of the life-span,39 slow growth,40 adiposity in early life,82 and reductions in competing organs86 are all adaptive responses to the costs of the large Homo brain. Kaplan, Lancaster, and Robson,39 for example, considered brain growth the key element of human life history, around which other components had been shaped. This appears to fit with more general life-history principles based on cross-species comparisons, in which brain size is associated with all other life-history variables. However, as Hawkes87 noted, the direction of these associations is still debated. Understanding why Homo developed a large brain is even more contentious.

Potts2 suggested that the larger brain, like other components of generalized biology, reduces the need to commit physiologically or anatomically to any specific niche. Effectively, the brain became a common hardware system on which multiple behavioral software systems could be run, according to ecological conditions.2 The question is, did other life-history traits indeed fit around increasing encephalization, or was encephalization merely one component of a suite of traits, each individually favored by increased ecological stress?

In this article, I have suggested that human life history has many features that enhance the capacity to tolerate ecological shocks, raising the possibility that large brains may have been less central to the evolution of Homo life history than is commonly assumed (Box 3). Importantly, organisms that tolerate stochastic environments need to be able to ride out harsh conditions and breed rapidly in good conditions. Shifts in life-history variables help meet both these requirements, thus raising the possibility that reduced adult mortality was not the only stress reshaping human life history. Equally, opportunistic rapid breeding with a cooperatively funded “conveyor belt” strategy may initially have evolved less to resolve high brain costs than to cluster reproductive effort between ecological shocks. Consistent with the arguments of Hawkes and Hrdy,36, 87 therefore, more generalized selection for slower growth, adiposity, and cooperative breeding arising from stochastic environments may have preceded and enabled the extreme encephalization of later Homo, through positive feedback mechanisms. Such issues merit further discussion. The aim here has been to put up a contrasting hypothesis for further evaluation. This perspective helps break new ground because, in focusing on adaptations to stochasticity, humans appear less unique and more consistent with other species facing similar ecological challenges.2, 20

Box 3. The Brain or Life History Under Selection?

Human brains are energetically costly, especially in early life,82 and should be favored only if the adaptive benefits outweigh the costs. Many potential ecological stresses favoring cognitive capacity have been proposed, including social competition and complex foraging activities.39, 96 If these stresses drove encephalization, other traits may have co-evolved in response. Recently, some have suggested that shifts in various other human traits, such as cooperative breeding and adiposity, may have preceded encephalization.36, 44, 87 These contrasting scenarios are illustrated in Figure 5.

5. Alternative scenarios for the evolution of encephalization. (a) The traditional cognitive capacity model assumes that social or skill-learning pressures favored larger brains and that other traits co-adapted to resolve the increased costs. (b) The alternative ecological stochasticity model assumes that life history was reorganized to address ecological perturbations and the resulting adaptations then enabled more energy to be diverted to brain growth.

Initially, human life history and physiology may have undergone reorganization to resolve ecological stochasticity. The adaptations may have become so effective that they not only addressed this stress, but also provided sufficient energy to allow an increase in brain size. From that point, the process could have proceeded by positive feedback; for example, slower growth, more cooperative breeding, and larger brain size could mutually reinforce each other.

Though humans and orang-utans may both have been exposed to ENSO, their different responses are likely to derive from the different spacing of food resources. Orang-utans appear to have sacrificed sociality in order to make subsistence returns viable at the level of the individual, and grow and reproduce at a very slow rate. Humans, in contrast, appear to have developed a more sophisticated “bet-hedging” response, increasing the production of offspring and using several strategies (slowed growth, cooperative care, and adiposity) to meet the resulting costs.

It is important to remember that Washburn's argument, that a suite of traits emerged simultaneously in human evolution,88 has been extensively criticized.87 If climate stochasticity did exert a profound and unifying effect on hominin evolution, as suggested here, it could still have done so incrementally, over an extended period, with adaptive change in one period building on changes and stresses in previous periods. The cumulative effect would be to make hominins less vulnerable to ecological catastrophes and better placed to benefit opportunistically from good conditions. These two fundamental traits may then have enabled the notable colonizing success, first of Homo erectus and then of Homo sapiens.75

What kind of climate cycles might have operated over the last 2-3 million years remains to be established. ENSO offers a valuable model in the contemporary world, but the actual pattern and magnitude of past climate perturbations, like their ecological effects, remain to be established. Improved methodologies are currently increasing the resolution with which past ecological change can be reconstructed. Importantly, ENSO cycles need not be drastic to generate the impacts on human biology described here. Seasonality and resource depletion represent continuous stresses for contemporary human foragers, who respond with nomadic behavior and physiological buffering systems such as adiposity, and demonstrate juvenile mortality rates of ∼40%, similar to those of contemporary apes.75, 89, 90 Irregular ENSO cycles acting on the early Homo niche would have superimposed additional stress for certain periods, potentially increasing mortality and delaying conception.

Extreme events may be important for understanding not only the emergence of hominin traits, but also how evolutionary adaptation occurred across generations. Human developmental plasticity is largely restricted to fetal life and infancy, and occurs under the transducing effect of maternal phenotype.19 The notion that each generation could be exposed to contrasting ecological states, with no systematic pattern, may help us understand why maternal phenotype exerts such a strong influence on the developmental trajectory of human offspring.19, 65 Some have argued that early in life offspring receive cues of future adult conditions.91 In contrast, I have argued that such a mechanism is implausible because stochastic environments do not demonstrate such auto-correlations over time.92 In such conditions, the cues received by offspring relate primarily to maternal phenotype, which buffers ecological perturbations.19, 65 Greater behavioral and physiological plasticity are central to the toleration of unpredictable environments in adult life, as highlighted in this review.21