R1.1. What is the mechanistic basis of the two-tiered model?

Kuzawa, Rosenbaum, and Gettler (Kuzawa et al.) argue that the two-tiered model is “mechanistically ambiguous and introduces unnecessary complexities.” They contend that most of the data reviewed in the target article could be more parsimoniously explained by a simple principle: Growth and puberty are energetically driven, but adult reproductive function is not. Thus, according to Kuzawa et al., energetic constraints are the only meaningful distinction between first- and second-tier processes.

First and foremost, although Kuzawa et al.’s simple principle can potentially explain why energetic stress and ambient cues to EM have distinct effects on LH traits, it cannot explain why these different sources of EM have countervailing effects on LH traits – the central premise of the two-tiered model.

Second, Kuzawa et al. question the mechanistic basis of the two-tiered model. As discussed in the target article, EM is communicated to the organism directly through biological pathways (e.g., energetic stress is signaled directly by gastrointestinal hormones; infectious diseases are signaled directly by immune responses) and indirectly through ambient cues in the external environment that are received and processed through sensory systems. First-tier developmental processes are driven by internal physiological states (e.g., energy reserves, immune function, parasite load, hormone concentrations) that result from a history of genetic, environmental, and developmental influences. First-tier conditions induce tradeoffs that divert energy away from growth and reproduction. At the level of the first tier, pathways from harsh ecological conditions (e.g., resource scarcity, infectious disease burden) to changes in physiological mechanisms of energy allocation (e.g., neuroendocrine changes that constrain physical growth or suppress ovarian function) do not require mediation through sensory systems that detect and encode information about the external environment.

By contrast, at the level of the second tier, ambient cues to EM are received and processed through sensory systems, which activate central nervous system processes (e.g., threat appraisal), which may in turn alter physiological mechanisms of energy allocation. Central nervous system processes, however, can influence reproductive behaviors (e.g., desired fertility, contraception choice, coital frequency) in ways that are not tightly constrained by such mechanisms. A defining feature of second-tier developmental processes is that they calibrate adult reproductive parameters through their effects on behavioral mediators (i.e., behavioral mediators are the intervening mechanism). By contrast, although first-tier developmental processes can affect behavioral mediators (e.g., immune system responses may activate the behavioral immune system; see Ackerman, Hill, & Murray, Reference Ackerman, Hill and Murray2018), the internal physiological states that define the first tier can alter mechanisms of energy allocation directly (without behavioral mediation). In total, first- and second-tier developmental processes are mechanistically distinct. Contrary to Kuzawa et al., this distinction is not adequately explained by the presence versus absence of energetic constraints.

R1.2. Does the two-tiered model err in its causal interpretation of mortality-fertility relations?

In his commentary, Volk questioned the causal interpretation of the findings reviewed in the target article:

This is not an either/or question. As stated in the target article, the answer is both: Just as reductions in mortality contribute to lower fertility, increases in mortality contribute to higher fertility. Micro-longitudinal data on reproductive change over the demographic transition in the Netherlands, Sweden, and Spain directly demonstrate this bidirectionality: Familial experiences of own-child survival (low mortality) played a central role in shifting reproduction toward longer interbirth intervals and lower total fertility, whereas experiences of own-child mortality had the opposite effect (target article, sect. 2.2.3). More generally, the target article reviews myriad studies demonstrating that women speed up pace of reproduction and increase fertility in the face of ambient cues to EM. Some examples include the contribution of malaria-mediated increases in child mortality to higher fertility (target article, sect. 2.2.6); reversals of the fertility transition in countries experiencing severe HIV epidemics (Gori et al., Reference Gori, Lupi, Manfredi and Sodini2020); faster pace of reproduction and increased fertility in response to mortality shocks caused by natural disasters and protracted armed conflicts in low-resource settings (target article, sect. 2.4.1); historical data linking childhood exposures to sibling mortality to earlier ages at marriage and first birth (target article, sect. 2.3.2); more frequent funeral attendance predicting shorter latency to pregnancy (Smith-Greenaway, Yeatman, & Chilungo, Reference Smith-Greenaway, Yeatman and Chilungo2022); physical proximity to a homicide predicting shorter latency to pregnancy (Weitzman, et al., Reference Weitzman, Barber, Heinze, Kusunoki and Zimmerman2023); and greater exposure within families (between sisters) to wartime activities predicting faster pace of reproduction and higher total fertility over the life course (Lynch et al., Reference Lynch, Lummaa, Briga, Chapman and Loehr2020).

R1.3. How does child versus prime-age adult mortality affect LH outcomes?

As presented in the target article, the two-tiered model was indefinite regarding the effects of ambient cues to child versus prime-age adult mortality on LH outcomes. We did offer the following provisional distinction:

We emphasized that this distinction was tentative, given inconsistencies in the data, and that further work was needed to more precisely delineate the evolved decision rules that guide different LH responses to ambient cues to EM in relation to child versus adult mortality risks. Likewise, Shenk and Varas Enriquez and Borgerhoff Mulder highlight the need for further theory and research in this area.

Jones contends that our tentative approach to the age-specificity of EM risk is a weakness of the two-tiered LH model. He argues that any theory of adaptive plasticity of LH traits needs to “account for age-specificity of vital rates,” particularly the high levels of juvenile relative to adult mortality in humans. In turn, Pepper was intrigued by the question of child versus adult mortality risks and took up the call for further work, conducting relevant data analyses. In this section, we respond to Jones’ critique and consider Pepper’s results in relation to other demographic work (including recent research reviewed by Shenk).

Based on a simple mathematical model of how age-specific mortality determines optimal reproductive effort (Charnov & Schaffer, Reference Charnov and Schaffer1973), Jones contends that the hypothesis that heightened juvenile mortality induces tradeoffs favoring offspring quantity over quality is wrong. According to Jones, Charnov and Shaffer’s formulation establishes that “high juvenile mortality relative to adults favors reduced reproductive effort.” However, this conclusion (which Charnov and Shaffer did not present as a general result) follows from the simplifying assumption that the age at first reproduction is fixed. In reality, reproductive effort is free to coevolve with reproductive timing, so that heightened juvenile mortality can favor earlier and higher reproductive effort. More generally, models that explicitly consider population density (omitted by Charnov and Shaffer) show that density-dependent effects on juvenile mortality tend to increase the benefits of concentrated reproductive effort (de Vries et al., Reference de Vries, Galipaud and Kokko2023; for further discussion, see sect. R2.2).

At an empirical level, little is known about the effects of ambient cues to child versus prime-age adult mortality on LH outcomes. Research examining the effects of child mortality rarely controls for adult mortality, and vice versa. That brings us to the between-country analyses conducted by Pepper, which supported the supposition that child mortality rates, compared with adult mortality rates (in women), are more strongly associated with total fertility (henceforth: Hypothesis 1), but did not support the supposition that adult mortality rates, compared with child mortality rates, are more strongly associated with early reproduction (henceforth: Hypothesis 2). In Pepper’s analyses, adult female mortality did not relate to either of these fertility outcomes, even after controlling for population measures of energetic condition (i.e., exposure to child and maternal malnutrition).

Pepper’s results contrast with past research on this topic, which (to our knowledge) includes only three studies (Lorentzen et al., Reference Lorentzen, McMillan and Wacziarg2008; Newmyer et al., Reference Newmyer, McAllister, Alam and Shenk2025; Ueyama & Yamauchi, Reference Ueyama and Yamauchi2009). Two important conclusions emerge from this previous research. First, taken together, these three studies afford mixed support for and against both Hypothesis 1 and Hypothesis 2 (for details, see Supplemental Material, sect. 4). Second, in contrast with Pepper, all three studies found large effects of prime-age adult mortality on LH outcomes (after controlling for child mortality). These large effects support the focus of the two-tiered model on behavioral mediators of adult reproductive parameters. Nonetheless, the small number of studies and inconsistent results make it difficult to evaluate either Hypothesis 1 or Hypothesis 2 at this time. More research is needed.

R1.4. Do evolved life-history adaptations bias parental investment toward offspring quantity over quality in response to heightened EM?

Based on his previous work on fitness elasticities, Jones criticized the idea that offspring quantity can be favored over offspring quality in human populations. Jones asserts that, because the fitness premium of juvenile survival (compared with adult survival) is so high, it almost never maximizes fitness to do anything that compromises juvenile survival – regardless of levels of EM. There are at least two problems with this argument. First, the fitness premium of juvenile survival cannot account for the wide variation in offspring number in natural fertility populations (Campbell & Wood, Reference Campbell, Wood, Diggory, Potts and Teper1988; Page et al., Reference Page, Ringen, Koster, Borgerhoff Mulder, Kramer, Shenk and Sear2024). This wide variation clearly indicates quantity-quality tradeoffs. Second, fitness elasticities only quantify the potential for selection on different vital rates, not their actual contribution to fitness. The latter also depends on constraints on those same vital rates, which is why elasticities alone can result in a distorted picture that fails to identify the strongest influences on fitness. For example, even if infant survival shows a high potential for selection, its actual contribution to fitness may end up being small if variation in this vital rate is highly constrained.

This is not just hypothetical. Davison and Gurven (Reference Davison and Gurven2021) calculated both the fitness elasticities and fitness contributions of age-specific survival and fertility rates in ten small-scale human societies. While elasticities showed the strongest force of selection on infant and child survival (in line with Jones), the actual fitness contribution of fertility was consistently greater than that of juvenile survival (infant and child survival combined). In total, in human populations, the impact of fertility can offset the impact of early survival, setting the stage for alternative resolutions of the quantity-quality tradeoff.

R1.5. Did selection favor an adaptation for contingently accelerating puberty in response to ambient cues to EM?

The two-tiered model proposes that, given adequate energetic conditions to support growth, evolved life-history adaptations should accelerate pubertal maturation in response to ambient cues to EM. Kuzawa et al. challenged this proposal. Consistent with the one-tiered model, they argue that the timing of puberty is tied to nutritional conditions, and that forager children’s nutritional status is often marginal with high rates of stunting. Accordingly, Kuzawa et al. contend that it was unlikely that adequate nutrition and high EM would have co-occurred with enough regularity for natural selection to have favored an adaptation for contingently accelerating puberty in this context.

Kuzawa et al.’s broad-stroke characterization of contemporary hunter-gatherers, and our ancestors, as living in poor quality environments without enough energetic resources to grow fast and mature quickly is inaccurate. High quality, repeat-observation, longitudinal demographic and anthropometric data for hunter-gatherers and other subsistence-level populations demonstrate that maturational trajectories are variable, not uniformly slow (Walker et al., Reference Walker, Gurven, Hill, Migliano, Chagnon, De Souza and Yamauchi2006; target article, sect. 2.1). Some hunter-gatherers and horticulturalists experience accelerated childhood growth and early pubertal maturation. The Savanna Pumé (Venezuela), Hiwi (Venezuela), Yanomamo (Venezuela), Baka (Cameroon), and Negritos (Philippines) face harsh conditions that are driven primarily by high parasite loads, infectious diseases, and/or violence (rather than malnutrition). Child survivorship (up to age 15) in these five populations ranges from 33% to 56% (Hackman & Kramer, Reference Hackman and Kramer2022; Walker et al., Reference Walker, Gurven, Hill, Migliano, Chagnon, De Souza and Yamauchi2006). Despite these harsh mortality conditions, Savanna Pumé, Hiwi, Baka, and Negritos all display accelerated patterns of childhood growth and maturation: Faster and greater linear growth across ages 3–10 followed by earlier ages at menarche than expected for their adult body size (Kramer & Greaves, Reference Kramer and Greaves2010; Walker et al., Reference Walker, Gurven, Hill, Migliano, Chagnon, De Souza and Yamauchi2006). Hiwi and Savanna Pumé girls have juvenile growth rates (Hiwi: 7.0 cm/yr; Pumé: 6.52 cm/yr) and ages at menarche (Hiwi: 12.6 yrs; Pumé: 12.9 yrs) that are similar to well-fed U.S. girls (Kramer, Reference Kramer2008; Kramer & Greaves, Reference Kramer and Greaves2010; Walker et al., Reference Walker, Gurven, Hill, Migliano, Chagnon, De Souza and Yamauchi2006). Contrary to Kuzawa et al., this accelerated development occurs despite limited energy available for growth (e.g., seasonal under-nutrition). In sum, the notion that growth and maturational timing are driven solely by energetics (Kuzawa et al.) does not adequately account for cross-cultural variation (for further discussion, see Supplemental Material, sects. 3 and 10).

In their analyses of 22 small-scale societies, Walker et al. (Reference Walker, Gurven, Hill, Migliano, Chagnon, De Souza and Yamauchi2006) found that age at menarche and age at first birth each occurred about 1 year earlier for every 10% decline in survivorship to age 15. Although accelerated development in this context could potentially be explained by mortality-driven selection favoring earlier/faster childhood growth and reproductive maturation (target article, sect. 2.1.3), accelerated development had to originate through developmental plasticity. As discussed below (sect. R1.8), because developmental processes generate variation on which selection can act, developmental change precedes and ultimately enables evolutionary change (West-Eberhard, Reference West-Eberhard2003). From this perspective, the pattern documented by Walker et al. (Reference Walker, Gurven, Hill, Migliano, Chagnon, De Souza and Yamauchi2006) – accelerated growth and reproductive maturation under conditions of high juvenile mortality – necessarily began as a developmentally plastic, conditional adaptation (even if selection ultimately acted on heritable components of the adaptation to establish accelerated growth and puberty through genetic changes). The question is whether accelerated development remains a developmentally plastic, conditional adaptation.

In high-income countries, where relevant variation in energetic conditions is largely neutral, meta-analyses have established that prospectively assessed childhood exposures to violence predict earlier pubertal development (d = −0.26) and its biological mediator, accelerated cellular ageing (d = −0.43) (Colich et al., Reference Colich, Rosen, Williams and McLaughlin2020; see also Ding et al., Reference Ding, Xu, Kondracki and Sun2024). In our view, the most parsimonious explanation for (a) violence-mediated acceleration of pubertal timing in high-resource settings, together with (b) data linking juvenile mortality to earlier reproductive development in small-scale societies, is that selection favored an adaptation for contingently accelerating growth and pubertal maturation in response to ambient cues to EM. In the two-tiered model, adequate energetics is a precondition for this adaptation to be expressed.

R1.6. Is the two-tiered model confounded by economic development?

Boucekkine, Bousmah, and Sahlin (Boucekkine et al.) argue that economic development is the main driving force behind the demographic transition and, specifically, the close coupling between declining mortality and declining fertility. They contend that “the effects of ambient cues to EM on LH strategies may not be considered independent of changes in economic development and their driving forces.” To start with, this criticism should be put into perspective: The two-tiered model was developed to explain how mortality-related experiences over development shape individual variation in human LH traits. The model was not intended to be a complete explanation of the demographic transition, though it is relevant to understanding aspects of the transition.

Economic models of fertility view children as consumer goods that provide utility (Becker, Reference Becker and Roberts1960). In highly competitive environments, parents optimize utility by producing few high-quality offspring at high levels of investment per child rather than many low-quality offspring at low levels of investment per child. Rising per capita income and associated technological changes and demand for human capital necessitate a high parental investment strategy (e.g., high investment in children’s human capital), which results in both lower total fertility and lower child mortality (e.g., Becker, Reference Becker1992).

We clearly acknowledge this economic explanation in the target article. That is why, in analyzing mortality-fertility relations, most between-country analyses (target article, sect. 2.2.5), and all of the within-country analyses (target article, Table 2), controlled for socioeconomic factors. In these multivariate analyses, better socioeconomic conditions and lower mortality each uniquely predicted declining fertility. Accounting for socioeconomic variation did not negate or confound the effects of mortality. Although these results are not inconsistent with economic models (Boucekkine et al.), the multivariate effects of economic development and declining mortality on quality-quantity tradeoffs are predicted by LH models: Resource-allocation strategies favoring offspring quality over quantity should maximize fitness under co-occurring conditions of low EM and high-resource competition – conditions in which high parental investment is costly to the parent but critical to offspring success (Kaplan & Lancaster, Reference Kaplan, Lancaster, Cronk, Chagnon and Irons2000).

The importance of low EM in this equation is illustrated by the paradox of Israeli demography: On the one hand, (a) Israelis have an exceptionally long life expectancy and experience high levels of economic opportunity and educational attainment; on the other, (b) they are immersed in ambient cues to EM. Compared with other Western countries (which experience [a] but not [b]), Israel maintains a high fertility rate that is well above replacement level (Sasson & Weinreb, Reference Sasson and Weinreb2024). Underscoring the potency of ambient cues to EM, Israel’s high fertility rate persists (and has even increased since the 1990s) despite high per capita GDP and increasing population density.

As recommended by Boucekkine et al., future research on LH strategies could benefit from moving beyond the current approach of controlling for socioeconomic variation. This could involve, for example, modeling interactions between EM and indicators of economic development (as in the Israeli paradox) and/or studying economic pathways through which declining EM contributes to lower fertility (e.g., changes from kin-based to non-kin-based social exchange in educational and work environments; Newson, Reference Newson2009). As a step in this direction, we discuss interactions between EM and resource competition in section R2.1.

R1.7. Are mortality-fertility relations genetically confounded?

Figueredo, Peñaherrera-Aguirre, and Hertler (Figueredo et al.) raise the issue of genetic confounding. They discuss the ostensibly high heritability of LH strategy – writing as if high heritability was a fact about LH strategy – without acknowledging that heritability coefficients are specific to the population from which they are calculated at a given period of time. Consider the heritability of three LH traits that are central to the two-tiered model: Age at first birth, adolescent fertility, and total fertility in women. In high-resource settings (i.e., industrialized and developed societies), the heritability of these traits has greatly increased over time.Footnote ¹ In a study of Danish twins born in birth cohorts ranging from 1870 to 1964 (used to calculate h ² for total fertility) and from 1945 to 1968 (used to calculate h ² for age at first birth and adolescent fertility), the heritability of each of these three traits increased from around 0% (in the oldest cohorts) to 40-59% (in the youngest cohorts) (Kohler, Rodgers, & Christensen, Reference Kohler, Rodgers and Christensen1999, Reference Kohler, Rodgers and Christensen2002). Shared environmental influences simultaneously decreased. These changes occurred as environments became more uniform over time, resulting in decreased variance in fertility (see Kohler et al., Reference Kohler, Rodgers and Christensen1999, Table 1). Of the variance that remained, more of it became attributable to variance in genes. This increasing genetic variance may reflect a societal context (without energetic constraints) in which women have a relatively broad range of life-course alternatives, increasing the pathways through which genetic influences can affect fertility outcomes (Kohler et al., Reference Kohler, Rodgers and Christensen1999, Reference Kohler, Rodgers and Christensen2002). In any case, when Figueredo et al. claim that LH strategy is highly heritable, that claim is specific to contemporary, wealthy, industrialized populations (for further discussion, see Supplemental Material, sect. 2).

Given the substantial heritability of LH traits in high-resource settings, we chose to focus the target article on small-scale human societies and low-and-middle-income countries (LMICs), wherein people often face significant energetic constraints on growth and reproduction. Our assumption was that small-scale societies and LMICs provide appropriate contexts for studying environmental regulation of human LH strategies. A study of the heritability of LH traits in a preindustrial Finnish population living under pre-demographic transition mortality and fertility conditions supports this assumption: Pace of reproduction (interbirth intervals; h ² = 29%) and total fertility (h ² = 31%) displayed moderate heritability, while age at first birth (h ² = 5%) and offspring survival (h ² = 0%) displayed low heritability (Pettay et al., Reference Pettay, Kruuk, Jokela and Lummaa2005). Thus, in this high-mortality, natural fertility population – where vital rates aligned with prevailing conditions over much of human history (see sect. R1.9) – there was ample room for variance in environmental conditions to shape life histories.

Nonetheless, we acknowledge that, in many cases, environmental correlations at the individual level are genetically confounded (e.g., Hart, Little, & van Bergen, Reference Hart, Little and van Bergen2021), as emphasized by Figueredo et al. To address possible genetic confounds, we devoted an entire section (target article, sect. 2.4) to analyzing natural experiments, whereby an environmental factor is altered (introduced or removed) by circumstances outside of the control of the affected people/communities. The impact of the altered environmental factor can then be evaluated. Natural experiments enabled testing of the effects of externally caused changes in mortality, providing evidence that mortality shocks caused by natural disasters and protracted armed conflicts in low-resource settings led to higher fertility (despite deteriorating energetic conditions that constrained childhood growth). Nonetheless, Figueredo et al. criticized the target article for focusing on environmental effects at the expense of genetic effects – despite offering no explanation for how genetic confounds could explain the countervailing effects of energetic stress and ambient cues to EM that are central to the two-tiered model.

R1.8. Is the ecological gambit plausible?

Borgstede contends that there is no a priori reason why individual-level adaptations (e.g., developmental adaptations to EM) should align with population-level adaptations (e.g., evolutionary adaptations to EM). Del Giudice (Reference Del Giudice2020) refers to the assumption that variation within species will align with variation between species as the ecological gambit. The gambit is that principles from LH theory, which were originally used to explain how selection pressures shape LH traits over evolutionary time, can inform our understanding of how developmental experiences shape LH traits within individual lifetimes.

West-Eberhard’s (Reference West-Eberhard2003) theory of developmental plasticity and evolution affords a potential mechanistic basis for the ecological gambit. Evolutionary change is dependent on developmental change (i.e., phenotypes must develop to be selected); thus, developmental and evolutionary changes often proceed hand-in-hand (Jablonka & Lamb Reference Jablonka and Lamb2014; West-Eberhard Reference West-Eberhard2003). From this perspective, developmental change is the horse that pulls the cart of evolutionary change.

Organisms are responsive to alterations in the conditions of their lives, whether those alterations originate from mutations or persistent changes in their environment. According to West-Eberhard’s (Reference West-Eberhard2003) model, selection favors individuals who respond in a plastic and functional manner to the new conditions and who preserve these changes through cross-generational transmission. These transmissible changes in developmental systems are then extended and adjusted by natural selection acting on the genetically heritable components of the systems.

In total, contra Borgstede, selection pressures in the environment tend to move evolution and development in the same direction in a mutually reinforcing manner. Organized developmental responses, whether genotypically or environmentally induced, precede and ultimately enable systematic (adaptive) evolutionary change, as developmental processes generate variation on which selection can act. Thus, the principles that govern the effects of EM on the evolution of LH strategies can inform hypotheses about the effects of EM on the development of LH strategies (as per the ecological gambit). Nevertheless, the ecological gambit is not guaranteed in any given case and should be tested rather than simply taken for granted (Del Giudice, Reference Del Giudice2020).

R1.9. Does higher fertility trade off against a shorter lifespan?

In his commentary, Međedović questioned the prevalence and importance of LH tradeoffs:

Tradeoffs between higher reproductive effort (faster pace of reproduction, higher fertility) and lower survival (faster senescence, reduced longevity) are a common assumption in LH theory (e.g., Takeshita, Reference Takeshita2024). In heterogeneous populations, however, such tradeoffs may be difficult to detect in comparisons between individuals who differ in physical condition, access to resources, social support, and related factors. This is because a person who is in good physical condition and has ready access to food, shelter, and a supportive kin network may be able to grow up faster, achieve larger adult size, live longer, have more children, and produce higher-quality offspring than another person who is in poor condition and has meager resources and little kin support. In preindustrial Finland, for example, higher offspring number traded off against lower offspring survival in resource-poor families but not resource-rich families (Gillespie, Russell, & Lummaa, Reference Gillespie, Russell and Lummaa2008). Such disparities often generate positive phenotypic correlations between people in LH traits that are in fact negatively correlated within persons (e.g., fertility versus longevity). Consequently, unless women’s health and socioeconomic conditions are controlled for, correlations between offspring number and female life expectancy in natural fertility populations do not reliably emerge (see Hurt, Ronsmans, & Thomas, Reference Hurt, Ronsmans and Thomas2006).

Equally important, genotypic as well as phenotypic correlations can be confounded by condition and resources. Specifically, if certain genetic factors affect an individual’s physical condition and/or ability to accrue resources, then the genetic correlation between fertility and longevity may be confounded (see Houle, Reference Houle1991; Wilson, Reference Wilson2014; Del Giudice, Reference Del Giudice2020). Although genetic correlations are not directly confounded by environmental factors – and thus can be informative, especially in the context of a mismatched environment – genetic correlations are not “pure” reflections of underlying tradeoffs and should not be uncritically interpreted as such.

Pulling these pieces together, a major study analyzed detailed LH data across multiple generations of a preindustrial Finnish population (18^th–19^th centuries), wherein social class differences were generally small and fertility and mortality rates were high (mean number of births per woman: 5.76; mean juvenile mortality rate: 31%; Pettay et al., Reference Pettay, Kruuk, Jokela and Lummaa2005). Consistent with the putative tradeoff between reproductive effort and survival, women who started reproducing at younger ages, or who reproduced at a faster pace, or who achieved high fertility before age 25, died at younger ages (Hayward, Nenko, & Lummaa, Reference Hayward, Nenko and Lummaa2015; Pettay et al., Reference Pettay, Kruuk, Jokela and Lummaa2005; analyses controlled for social class). Contra Međedović, in this natural fertility population (without advanced medical care or modern contraception), both phenotypic and genotypic correlations between fertility and longevity were negative, not positive (Pettay et al., Reference Pettay, Kruuk, Jokela and Lummaa2005; see Blomquist, Reference Blomquist2009, for a similar finding in rhesus macaques). In total, these data suggest that “the greatest challenge for LH theory” (Međedović) may not be applicable to human LH tradeoffs prior to the demographic transition.

R2.1. Population density should be included in a comprehensive model of human LH variation

Several commentators discussed the importance of density-dependent regulation of LH traits (Komyaginskaya, Gallyamova, & Grigoryev [Komyaginskaya et al.]; Li & Lim; Lu, Zhou, Yang, Zhu, & Chang [Lu et al.]; Volk). Higher population density is strongly associated with indicators of human development, including greater urbanization, higher per capita GDP, more years of education, and lower rates of child mortality (de la Croix & Gobbi, Reference De la Croix and Gobbi2017). Consistent with this pattern of covariation, density-dependent LH models stipulate that conditions of high population density favor higher competitive ability – to successfully control limited resources in saturated environments – and resilience to the detrimental effects of greater intra-specific competition (Ellis et al., Reference Ellis, Figueredo, Brumbach and Schlomer2009; MacArthur & Wilson, Reference Macarthur and Wilson1967; Wright et al., Reference Wright, Bolstad, Araya-Ajoy and Dingemanse2019). As reviewed by Li and Lim, applications of this density-dependent model to human development support the hypothesis that greater population density favors greater investment in embodied capital (e.g., skills and knowledge), slower pace of reproduction, lower offspring number, and higher investment in offspring quality (Rotella et al., Reference Rotella, Varnum, Sng and Grossmann2021; Sng et al., Reference Sng, Neuberg, Varnum and Kenrick2017). This shift toward a slower LH strategy is specifically predicted in high-density environments that are relatively stable/predictable and thus can support longer-term investments in future reproduction (Rotella et al., Reference Rotella, Varnum, Sng and Grossmann2021).

Different processes operate when high-density environments are linked to greater energetic deprivation and juvenile mortality (see Lu et al.), as documented in Walker and Hamilton’s (Reference Walker and Hamilton2008) analysis of density-dependent effects on mortality, growth, and reproductive timing in small-scale human societies. As per the one-tiered model, higher population density in small-scale societies is associated with increases in malnutrition and infectious disease, which slow down growth and pubertal development and delay the onset of reproduction (Walker & Hamilton, Reference Walker and Hamilton2008). Furthermore, contrary to Jones’ assertion that “human populations manage to avoid density-dependent effects on vital rates,” juvenile mortality in small-scale societies increases with higher population density (Walker & Hamilton, Reference Walker and Hamilton2008; Wood & Smouse, Reference Wood and Smouse1982; Waguespack, Reference Waguespack2002). In turn, heightened levels of juvenile mortality – a powerful ambient cue to EM (target article, Table 2) – alters behavioral mediators in ways that promote a faster pace of reproduction and higher offspring number (target article, sect. 1.3). In total, density-dependent competition may regulate both first-tier and second-tier developmental processes, with countervailing effects: Slower growth and later reproductive development resulting from energetic constraints and infectious diseases, but a faster pace of reproduction and higher offspring number resulting (indirectly through behavioral mediators) from ambient cues to EM (in this case, community-level juvenile mortality). This conclusion is consistent with evidence showing that the robust association between higher population density and lower fertility is suppressed under harsh environmental conditions (Rotella et al., Reference Rotella, Varnum, Sng and Grossmann2021).

What is needed is research comparing the effects of population density, socioeconomic factors, and mortality rates (and their interactions) on LH outcomes. Komyaginskaya et al. attempted to address this research question in their empirical analyses. Unfortunately, their data are difficult to interpret because (a) they performed a series of univariate analyses, rather than testing a larger multivariate model, and (b) their dependent variable merged mortality and fertility indicators into a single composite measure. De la Croix and Gobbi (Reference De la Croix and Gobbi2017) also addressed this research question, conducting within-country analyses (change over time) that examined the effects of population density, socioeconomic factors, and infant mortality on fertility in 24,769 communities across 44 LMICs. They found that each of these three community-level environmental factors accounted for unique variation in fertility, including evidence of causal relations based on the temporal ordering of change within populations over time. Whereas both greater population density and better socioeconomic conditions predicted lower fertility, higher infant mortality had the opposite effect. In total, we concur with multiple commentators (Komyaginskaya et al.; Li & Lim; Lu et al.; Volk) that population density regulates LH strategies and should be included in a comprehensive model of human LH variation.

R2.2. Formal modeling could augment the two-tiered LH model

Several commentators (Borgstede; Međedović; Schlomer; Varas Enriquez & Borgerhoff Mulder) argue for the need to build and test formal mathematical models. Although the two-tiered model is supported by extensive empirical data, our analysis of EM aimed for simplicity. We could readily increase the complexity of the model by adding new parameters. In addition to incorporating density dependence (sect. R2.1), future research could formally model how different parameters of EM (e.g., chronicity, severity, predictability, timing, and their interactions) affect the costs and benefits of different LH tradeoffs.

As noted by Borgstede, such questions are too complex for verbal models (cf. Shenk, for a discussion of the timing of mortality exposures vis-à-vis different reproductive outcomes). Formalizing the parameters of EM as computational models and using simulations to assess the consequences (fitness costs and benefits) of different LH traits across a wide range of these parameters could enrich and refine the two-tiered model. This process could help to advance our understanding of which properties of EM are important for what outcomes. Most critically, by making assumptions explicit and analyzing their consequences, formal models can potentially distinguish between competing causal explanations (Stearns & Rodrigues, Reference Stearns and Rodrigues2020), such as the competing explanatory and predictive frameworks instantiated in the one-tiered versus two-tiered models.

Despite these strengths, formal models have limitations. In theoretical biology, formal mathematical models have established complex interactions between EM, density-dependent competition, and the evolution of life histories (de Vries et al., 2023). Despite this progress, such models may be limited by reliance on simplifying assumptions that do not hold in the real world. Consider Borgstede’s summary of this literature:

Although this conclusion is mathematically valid, it is misleading because it assumes an exponentially growing population with no density-dependent regulation; however, in an exponentially growing population, density-dependent competition invariably occurs over time. When assumptions are modified to allow for density-dependent competition (adding realism to the model), and that competition disproportionately affects juveniles, then higher EM again selects for earlier reproduction (de Vries et al., 2023). These subtleties are important because predictions derived from formal models are only meaningful when based on assumptions that hold in the real world (Stearns & Rodrigues, Reference Stearns and Rodrigues2020). As noted above (sect. R2.1), in small-scale human societies, juvenile mortality increases with higher population density. If a formal model examining the impact of EM on the evolution of age at first reproduction in humans either omitted density-dependent regulation or modeled it incorrectly (e.g., specifying age-independent rather than juvenile-specific density dependence), the analysis would erroneously conclude that higher EM does not select for earlier reproductive timing (see de Vries et al., 2023). In sum, we agree with various commentators (Borgstede; Međedović; Schlomer; Varas Enriquez & Borgerhoff Mulder) that tests of empirical hypotheses drawn from LH theory could benefit from formal modeling – with the caveat that such models are based on realistic assumptions (for further discussion, see Supplemental Material, sect. 5).

R2.3. Should we expand the concept of ambient cues to EM: Prospects and pitfalls

The target article focused on ambient cues to EM: Developmental experiences and environmental exposures that (1) are received and processed through sensory systems and (2) signal premature death or heightened risk of death among individuals in one’s local environment (e.g., funeral attendance, own-child mortality, the prevalence of infant or child deaths among immediate social network or community members, physical proximity to a homicide). In short, we focused on ambient cues that directly indexed mortality-related experiences. Several commentators challenged the completeness of this approach (Dunin & van Buren; Maranges & Timbs; Papke, Del Toro, & Klimes-Dougan [Papke et al.]; Sasson; Sheppard; Szepsenwol) or argued that mortality risks (including energetic shortfalls) are buffered by integration into cooperative social networks (Gonzalez & Lee).

Many cues do not directly index mortality-related experiences but are probabilistically linked to such experiences (e.g., poverty, high crime rates, father absence, harsh parenting, identity-based discrimination, social disconnection). For example, identity-based discrimination in sexual and gender minority populations (Papke et al.), and social disconnection in adolescents (Gonzalez & Lee), may act as ambient cues to EM. Such indirect mortality cues (proxy indicators) could, in theory, activate the same behavioral mediators of LH strategies as do more direct indicators of mortality risk. This raises the question of cue reliability (Frankenhuis, Nettle, & Dall, Reference Frankenhuis, Nettle and Dall2019): The extent to which an experience or event provides information about a current or future state of the world (in this case, premature death).

Cue reliability of proxy indicators is an important theoretical and empirical question, as evolved neurobiological mechanisms that guide LH responses to EM should be discriminant – designed to detect and respond to cues that exceeded a reliability threshold over our natural selective history. Presuming adequate cue reliability, assessment of indirect cues to EM may be a methodological necessity for researchers working in high-resource settings, where participants may have few direct mortality-related experiences but still face other significant adversities. Early instantiations of the developmental hypothesis (e.g., Belsky, Steinberg, & Draper, Reference Belsky, Steinberg and Draper1991; Ellis, Reference Ellis2004) and many empirical tests of the model (e.g., Ellis et al., Reference Ellis, Bates, Dodge, Fergusson, John Horwood, Pettit and Woodward2003; James et al., Reference James, Ellis, Schlomer and Garber2012; Nettle, Coall, & Dickins, Reference Nettle, Coall and Dickins2011) employ proxy indicators of EM (but see Maranges & Timbs for a discussion of how such indicators can become a runaway train).

R2.4. Revising the concept of behavioral mediators: Prospects and pitfalls

As discussed in the target article, ambient cues to EM affect key behavioral mediators of adult reproductive function. For researchers working in this area, the choice of behavioral mediators raises some of the same theoretical issues as the choice of (direct vs. indirect) EM indicators. Chen criticized the target article for neglecting many potentially important mediators of EM. Although the target article discusses a plethora of behavioral mediators and clearly articulates their centrality to the two-tiered model – acting as pathways through which exposures to ambient cues to EM promote a faster pace of reproduction and higher offspring number – our discussion of mediators was circumscribed: We only focused on mediators that were empirically linked to mortality exposures. Chen advocates for considering a broader array of mediators (e.g., shyness, insecure attachment, sociability; see also Dunin & van Buren). The challenge with this approach is that this broader array includes mediators with uncertain or absent connections to mortality-related experiences. That complicates interpretation and takes us outside of the domain of the two-tiered model (for further discussion, see Supplemental Material, sect. 8). Nonetheless, Chen’s nuanced discussion of mediators (e.g., avoidant attachment amplifying the salience of energetic stress) nicely extends a key element of the two-tiered model.

Smith-Greenaway proposes that we should reconceptualize the idea of mediators in the two-tiered model. She questions the notion that people intentionally upregulate fertility, or mediators of fertility, in response to ambient cues to EM. Smith-Greenaway instead argues that mortality-related experiences (e.g., funeral attendance, the murder of a family member) impose such a burden of grief that it disrupts women’s ability to avoid pregnancy. Thus, in contrast to the two-tiered model, Smith-Greenaway focuses on impairment rather than adaptation to mortality-related experiences (cf. Sasson, who discusses bereavement within the context of the two-tiered model; see also Supplemental Material, sect. 6).

On the one hand, intense bereavement may help to explain some of the data reviewed in the target article, such as the disruptive effects of protracted warfare or natural disasters, or the link between deaths of family members or close social network members and shorter latencies to pregnancy. On the other hand, intense bereavement is unlikely to explain other data reviewed in the target article, such as: (a) the effects of childhood exposures to sibling mortality (target article, sect. 2.3.2) or armed conflict (Madsen & Finlay, Reference Madsen and Finlay2019; Međedović et al., Reference Međedović, Karić, Kostić and Kovačević2025), where faster LH strategies emerge years after the fact; (b) within-family differences between sisters who, as adolescents, were differentially exposed to war and subsequently displayed divergent LH strategies over the life course (Lynch et al., Reference Lynch, Lummaa, Briga, Chapman and Loehr2020); or (c) the effects of community- or regional-level mortality on fertility (target article, Table 2), where heightened levels of premature death occur among people who, in many cases, are only remotely associated with the participants. Most importantly, there is clear evidence that ambient cues to EM alter behavioral mediators of adult reproductive parameters in ways that promote faster pace of reproduction and higher offspring number (target article, sect. 1.3.2). In total, although bereavements may disrupt family planning, there is strong evidence that individuals detect and encode the mortality environment around them in ways that calibrate their life histories.

R2.5. The two-tiered LH model versus internal/external predictive adaptive response models: convergence and divergence

Several commentators (Coall, Sear, Kathigesu, & Chisholm [Coall et al.]; Lu et al.; Međedović) drew parallels between the two-tiered model and internal/external PAR models. Early life stress may cause damage to the soma (i.e., erode phenotypic condition in a manner that reduces health and longevity), and the damaged soma itself may function to accelerate the timing or pace of reproduction (Nettle, Frankenhuis, & Rickard, Reference Nettle, Frankenhuis and Rickard2013; Rickard, Frankenhuis, & Nettle, Reference Rickard, Frankenhuis and Nettle2014). In this view, developmental mechanisms respond to the individual’s compromised internal/physiological state (with that state acting as a bioassay of the person’s health and nutritional status; see Coall et al.) and not to probabilistic cues about its future environment. This proposed developmental trajectory has been referred to as an internal PAR (Nettle et al., Reference Nettle, Frankenhuis and Rickard2013); it converges with first-tier developmental processes insofar as both are driven by internal physiological states (see sect. R1.1). By contrast, LH researchers testing the developmental hypothesis (target article, sect. 1) generally assume that early adversity carries predictive information about the future state of the environment (e.g., danger and consequent high mortality), and that development of a faster LH strategy in this context functions to match the individual to this expected harsh external state. This hypothesized developmental trajectory has been referred to as an external PAR (Nettle et al., Reference Nettle, Frankenhuis and Rickard2013) and is regulated by second-tier developmental processes.

Lu et al. astutely observed a discrepancy between the predictions of the two-tiered and internal PAR models. Although both models conceptualize the organism’s internal somatic state as guiding adaptive developmental plasticity, the internal PAR model predicts that both compromised internal states and ambient cues to EM regulate development toward faster LH strategies, whereas the two-tiered model postulates that energetic stress (experienced internally as an altered physiological state) and ambient cues to EM have countervailing effects on the development of LH strategies. For example, contrary to the internal PAR model, the two-tiered model predicts that first-tier conditions (e.g., compromised metabolic condition, acute inflammation) guide resource-allocation tradeoffs toward slower growth and later puberty.

A study of the British Birth Cohort (Waynforth, Reference Waynforth2012) may be relevant to distinguishing between the kinds of internal states that guide internal PARs versus first-tier developmental processes. The study assessed childhood adversity (i.e., family socioeconomic adversity, father absence) and chronic health conditions known to reduce life expectancy at age 10. Each of these variables independently predicted earlier timing of first reproduction. However, childhood adversity and chronic health conditions were uncorrelated. That is, chronic health conditions in this high-resource setting signaled a compromised internal/physiological state that was independent of low-resource availability. Thus, chronically ill participants were unlikely to have experienced the kinds of energetic or pathogen stress that constrain physical growth and delay pubertal maturation.

Putting these pieces together, the internal PAR model proposes that compromised internal physiological states signaling a truncated life span (e.g., oxidative stress, telomere shortening) accelerate reproductive timetables. But that may only apply to a subset of compromised internal states (henceforth: Subset A). Internal states that constitute first-tier conditions (henceforth: Subset B), such as low body fat stores signaling energetic stress or high concentrations of C-reactive protein (acute inflammation) signaling infection, may have the opposite effect of Subset A. In total, Subset B but not Subset A can be expected to induce tradeoffs that divert energy away from growth and reproduction. Conversely, Subset A but not Subset B can be expected to induce tradeoffs that favor current over future reproduction. Future research is urgently needed to determine how and when these different stress-mediated internal/physiological states accelerate versus decelerate reproductive maturation and timing. Elucidating this distinction would have meaningful implications for science and policy.

R2.6. Sequential ordering of first-tier versus second-tier developmental processes

Boyd-Frenkel, Choi, and Sng (Boyd-Frenkel et al.) raise questions about the sequential ordering of first-tier versus second-tier developmental processes:

Boyd-Frenkel et al. also question whether the effects of ambient cues to EM (second-tier effects) follow a strictly linear pattern.

Mortality shocks may initially promote a faster pace of reproduction and higher offspring number through behavioral mediators of adult reproductive function. However, over time, these behaviorally mediated effects could be swamped by accumulating levels of energetic deprivation and somatic depletion that are severe enough to suppress reproduction. This apparently occurred over the course of the twentieth century in Dominica (target article, sect. 2.3.2). The two-tiered model proposes that – without meaningful energetic constraints – ambient cues to EM will have linear (positive) effects on the pace of reproduction and offspring number. However, because energetic stress and ambient cues to EM have opposing effects on adult life histories, the two tiers may jointly result in an inverted u-shaped curve, where periods of moderately high EM promote faster pace of reproduction and higher offspring number, relative to periods of either (a) low EM or (b) high EM imposed by first-tier conditions, as per the Dominica case study (Quinlan, Reference Quinlan2010). Furthermore, when first-tier conditions (e.g., high pathogen stress, famine) cause high levels of EM, parental care decreases (Quinlan, Reference Quinlan2007), indicating that effort allocated toward increasing offspring quality in this context only weakly mitigates offspring mortality. In total, high EM imposed by first-tier conditions may bias allocation of resources toward self-preservation (i.e., somatic effort), resulting in tradeoffs that decrease reproductive effort (including reductions in both offspring quantity and quality).

R3.1. COVID-19

One potential counterexample – wherein higher mortality ostensibly leads to lower fertility – involves COVID-19. Sun argues that, in more economically developed societies, perceived environmental insecurity or instability, such as the COVID-19 pandemic or social unrest, operate as important cues to EM. Contrary to the two-tiered model, Sun suggests that perceived environmental insecurity or instability guides reproductive decision-making toward slower LH strategies. In support of this position, Sun (and Volk) note that the COVID-19 pandemic has generally resulted in lower fertility. However, the impact of the COVID-19 pandemic on fertility is currently an unsettled question (e.g., Nitsche & Wilde, Reference Nitsche and Wilde2024; Sobotka et al., Reference Sobotka, Zeman, Jasilioniene, Winkler-Dworak, Brzozowska, Alustiza-Galarza and Jdanov2024; Xiao, Xin, & Wang, Reference Xiao, Xin and Wang2024), with initial data suggesting “an unexpectedly incoherent and heterogeneous response” (Nitsche & Wilde, Reference Nitsche and Wilde2024, p. 9). This incoherence is reminiscent of early research examining the fertility impact of the HIV/AIDs epidemic in sub-Saharan Africa, which also produced mixed results (e.g., Castro, Behrman, & Kohler, Reference Castro, Behrman and Kohler2015; Kalemli-Ozcan et al., Reference Kalemli-Ozcan2012; Young, Reference Young2007).

What explains the variable and unsettled nature of these initial fertility data? An issue with early (i.e., temporally proximal) assessments of the effects of COVID-19 and HIV/AIDs is that they do not take into account the approximately 10-year lagged fertility response to changes in mortality (e.g., Angeles, Reference Angeles2010). Longer-term demographic analyses have now shown that mortality shocks caused by the HIV/AIDS epidemic generally promoted higher fertility overall across sub-Saharan Africa (Chin & Wilson, Reference Chin and Wilson2018; Gori et al., Reference Gori, Lupi, Manfredi and Sodini2020). This necessitates a longer-term view of mortality-fertility relations; it is simply too soon to draw meaningful conclusions about the fertility impact of the COVID-19 pandemic.

R3.2. Cultural factors

It is well known that cultural factors can influence, or even foreclose, reproduction (e.g., Shakers adherence to celibacy). Both Zhong and Sun argue for the primacy of cultural factors in regulating mortality-fertility relations. Sun proposes that sociocultural factors, such as parental leave, education, and economic opportunity, may decouple mortality-fertility relations and override LH tradeoffs (e.g., leading people who have experienced heightened cues to EM and uncertainty to have fewer children). As reviewed in the target article, the empirical record does not support this proposal (see also the Israeli demographic paradox [sect. R1.6]).

Zhong claims that The Song Dynasty presents a counterexample to the mortality-fertility nexus. According to Zhong, the Song Dynasty ostensibly combined high levels of EM (resulting from warfare) with economic prosperity, active fertility restriction, and slow population growth or even population decline. This claim cannot be evaluated, however, because Zhong does not review data at the level of individual variation in mortality-related experiences (i.e., we do not know if people in the Song era who had more exposure to ambient cues to EM restricted their fertility more [or less] than peers who had less exposure). Nevertheless, this apparent anomaly (high mortality but low fertility) seems unlikely because – contrary to Zhong – the eleventh and twelfth centuries in China were apparently characterized by substantial population growth and may have been the time period when the Chinese population broke the record of 100 million (Zhao, Reference Zhao1997).