1 Endocrinology, Energetics, and Human Life History: A Synthetic Model 2 3 Peter T. Ellison 4 Department of Human Evolutionary Biology 5 Harvard University 6 7 8 9 Abstract 10 11 12 Human life histories are shaped by the allocation of metabolic energy to 13 competing physiological domains. A model framework of the pathways of energy 14 allocation is described and hormonal regulators of allocation along the pathways of the 15 framework are discussed in the light of evidence from field studies of the endocrinology 16 of human energetics. The framework is then used to generate simple models of two 17 important life history transitions in humans, puberty and the postpartum return to full 18 fecundity in females. The results of the models correspond very closely to observations 19 made in the field. 20 21 22

23 Introduction 24 25 The essence of life can be defined as “metabolism in the service of 26 reproduction.” Organisms capture energy from the environment and turn it into more 27 organisms. Natural selection has favored those variations that perform this task more 28 reliably and more efficiently than their competitors. But the route from energy capture to 29 reproduction can be complex. Even in the simplest organisms some captured energy 30 must be devoted to growth and maintenance of the organism as well as to reproduction. 31 Reproduction without growth would rapidly lead to smaller and smaller organisms until 32 the size limits of viability were reached. Investment in maintenance leads to increased 33 survivorship and opportunities to continue reproducing. The partitioning of available 34 energy among these non-overlapping categories gives rise to the patterning of life 35 histories -- variation in age-specific probabilities of mortality and fertility, trajectories of 36 growth, rates of senescence, and other aspects of phenotype that are only manifest in a 37 diachronic view, in the way an organism’s life unfolds rather than in its state at any 38 given time. 39 40 Life history theory emphasizes trade-offs and how optimal energy allocation 41 varies with age and environmental circumstances. But this body of theory often leaves 42 unspecified the physiological mechanisms that govern and regulate those trade-offs. 43 The field of human reproductive ecology has emerged out of an effort to understand 44 those mechanisms as they relate to the regulation of reproductive effort in particular

45 (Ellison, 2003a, 2009). More recent efforts to illuminate the mechanisms governing 46 trade-offs of investment in growth and immune function are helping to further advance 47 an integration of the physiological mechanisms of energy allocation decisions with the 48 conceptual framework of life history theory (Flatt and Heyland, 2011). The focus of this 49 paper will be a consideration of the ways the endocrine system helps to regulate 50 metabolic energy allocation to generate the structure of human life history. 51 52 The endocrine system is one of the three major systems of integration of 53 vertebrate physiology based on molecular communication, the other two being the 54 nervous system and the immune system. The power of the endocrine system as a 55 physiological integrator lies in two properties: (i) the diffuse nature of its communication, 56 reaching all cells of the organism, and (ii) its ability to regulate both cellular activity and 57 gene expression. Together these properties position the endocrine system to regulate 58 energy allocation in ways that integrate with developmental and life history changes. 59 60 There are, however, two frequent biases in endocrinology that we should be 61 particularly aware of in considering the regulation of metabolic energy allocation: the 62 “top down bias” and the “newcomer bias.” The top down bias is manifest in a 63 predilection for assuming that the brain is in charge of the body. In many areas this is 64 true, but it is not an absolute hierarchy of regulation. Elsewhere (Ellison, 2009) I have 65 noted that molecular communication can be classified by the channels through which it 66 flows into (a) central-nervous-system (CNS)-to-soma, (b) soma-to-CNS, (c) soma-to-

67 soma, and, (d) CNS-to-CNS. Behavioral endocrinology is dominated by soma-to-CNS 68 communication, coordinating behavior with the physiological state of the organism, 69 particularly as regards the regulation of reproductive effort. The most potent 70 messengers along this pathway are those that easily cross the blood-brain barrier, 71 especially steroids, as well as peptides that gain access to the basal hypothalamus. 72 CNS-to-soma communication primarily flows through the hypothalamic-pituitary 73 “transducer”, integrating sensory and other information from basal ganglia and higher 74 cortical areas in the regulation of peripheral organs and endocrine glands. Larger 75 protein molecules are typical messengers along this pathway. Both the soma-to-CNS 76 and CNS-to-soma channels can have important effects on the allocation of metabolic 77 energy. But the primary conduit for messages regulating energy allocation is the soma78 to-soma channel. This pathway includes messengers of all chemical types, including 79 steroids (both gonadal and adrenal), thyroid hormones, proteins (including pancreatic 80 hormones), and peptides (including gut hormones and adipokines). In large part, as will 81 be described, the action of messages flowing through the CNS-to-soma pathway on 82 energy allocation is achieved by modifying the action of messages in the soma-to-soma 83 pathway. 84 85 The newcomer bias is manifested in a natural fascination with newly discovered 86 or described messenger molecules and the desire to see them as particularly important 87 in their effects. This can often be the case, but should not cause us to neglect the 88 central roles that are often played by those messengers who have long featured in our

89 understanding of endocrine physiology. Many of these, such as steroids, thyroid 90 hormones, pancreatic and gut hormones, also feature prominently in the soma-to-soma 91 channel. 92 93 Keeping these biases in mind, we can consider the major features of the 94 endocrine regulation of energy allocation decisions and their impact on human life 95 history. The framework described here will necessarily be an over-simplification, like all 96 models. Yet hopefully it will capture enough of the important features of the system to 97 be heuristic, in organizing and synthesizing information, in illuminating complex 98 interactions and processes, and in generating hypotheses. Where possible, we will pay 99 attention to evidence from and applications to field studies of human physiology. 100 101 The endocrine framework of human energetics 102 103 The basic framework of human energetics, the flow of energy through the 104 organism, is represented by the diagram in Figure 1 where the pathways indicated by 105 specific arrows are associated with the hormonal regulators shown in Figure 2. As a 106 model, not all the pathways of physiological integration and regulation that affect 107 energetics are included, nor all the potential endocrine signals involved. Those that are 108 included are considered to be major pathways and regulators which account for major 109 aspects of energy allocation and its life history effects. 110

111 At the center of the framework is “available metabolic energy,” available for 112 immediate allocation to any one of a number of non-overlapping categories. Inputs to 113 available metabolic energy come either from direct intake, or from stores, filtered by the 114 regulation of ATP production. The major allocation categories for available metabolic 115 energy are identified as storage, activity, anabolism, and maintenance. Anabolic sub116 categories include growth and reproduction. The parts of this framework will be 117 considered in turn. 118 119 120 Regulation of energy balance 121 122 Energy balance refers to the difference between energy intake through nutrition 123 and energy expenditure through all metabolic pathways. Of the various pathways of 124 energy expenditure, physical activity can be separated out as distinct from those that 125 contribute to resting metabolic rate. The net of energy intake and expenditure in activity 126 can be considered as the contribution to available metabolic energy from the 127 environment. 128 129 Energy intake is not tightly regulated by the endocrine system, since it depends 130 greatly on environmental factors. Appetite regulation, however, is strongly affected by 131 endocrine signals (Badman and Flier, 2005; Blundell et al., 2015a; Blundell et al., 132 2015b; Crespo et al., 2014; Meier and Gressner, 2004; Schwartz, 2000; Schwartz et al.,

133 2000; Webber et al., 2015). Two signaling molecules that will be considered in this 134 model are leptin and ghrelin. Both are peptides, principally produced in the soma but 135 capable of some penetration of the blood-brain barrier. Leptin is primarily produced by 136 adipocytes, ghrelin by the gut. Both carry information to the hypothalamus that can 137 affect appetite. Leptin, a tonic hormone reflecting the mass of adipocytes, is associated 138 with reduced hunger when circulating levels are high and increased hunger when levels 139 are low (Schwartz et al., 2000). Ghrelin, an episodic hormone which reflects short-term 140 status of gut contents, with high levels occurring when the gut is relatively empty for an 141 extended period, stimulates hunger when its levels are high (Pinkney, 2014). These 142 orexigenic effects are mediated by and coordinated with other neural signaling, primarily 143 in the ventromedial hypothalamus (Webber et al., 2015). 144 145 Energy expenditure in activity is not under tight hormonal control in humans, but 146 in rodent models, lowered leptin levels were early observed to correlate with increased 147 levels of physical activity, perhaps reflecting a stimulation of the motivation to forage for 148 food (Pelleymounter et al., 1995; Wolf, 1996). In many rodents, food foraging exposes 149 the animal to significant predation risk. Appetite regulation may serve to help regulate 150 foraging effort in a way that balances risks versus benefits. There is no evidence at this 151 time that foraging effort in human hunter-gatherer societies (the evolutionarily salient 152 human subsistence pattern) or any other ecological context is related to hormonal 153 appetite regulation. However, there is clinical evidence that exercise may affect 154 appetite through the mediation of appetite regulating hormones (Blundell et al., 2015b).

155 156 Studies of human leptin and ghrelin levels in the field have produced some 157 notable results, particularly highlighting population variation in average levels 158 (Bribiescas, 2001, 2005; Bribiescas and Hickey, 2006; Kuzawa et al., 2007; Miller et al., 159 2006; Munch-Andersen et al., 2013; Sharrock et al., 2008; Tanaka et al., 2005). In 160 general, subjects in non-western populations and populations engaged in subsistence 161 economies have significantly lower levels of both appetite regulating hormones than in 162 western, developed societies. This is true for leptin even after correcting for fat mass. 163 At the least this suggests that caution needs to be exercised in interpreting the 164 significance of absolute levels of these hormones. It is likely that the set-points for 165 appetite regulation may be influenced by developmental factors (Sharrock et al., 2008) 166 and may reflect population differences in overall energy budgets. 167 168 169 Regulation of energy availability 170 171 a. Release of oxidizable substrates 172 173 Oxidizable substrates (principally carbohydrates and fatty acids) are released 174 from stores into the blood stream under the regulation of several different hormones. 175 Among the most important regulators of energy substrate release in humans are 176 cortisol, epinephrine, and glucagon. Epinephrine and glucagon are relatively short-term

177 regulators, involved in defending blood sugar homeostasis against the vagaries of 178 intake and expenditure on a time scale of minutes to hours. Because of their short-term 179 effects, these hormones are often involved in behavioral responses, including “fight or 180 flight” scenarios. However, they are not directly implicated in the longer term regulation 181 of energy substrate release that is typically involved in life history strategies and 182 transitions. 183 184 Cortisol, on the other hand, is involved in longer term regulation of energy 185 substrate availability. Cortisol has been associated with responses to psychosocial 186 stress, to the extent that it is frequently referred to as a “stress” hormone. However, it 187 can be misleading to label cortisol in this way, since psychosocial stress is only one 188 potential trigger for its release. Other than categorizing cortisol by one of the multiple 189 factors the can cause its release, it would be better to categorize it by its downstream 190 effects as a “metabolic” hormone. Cortisol’s principal actions involve the stimulation of 191 lipolysis to release fatty acids from adipose stores and the antagonism of 192 gluconeogenesis, resulting in an increase in available oxidizable substrates (Widmaier 193 et al., 2013). Although cortisol is elevated as a consequence of short-term stresses, it 194 rapidly returns to baseline when those stresses are removed. Chronic energy 195 demands, as can occur with infection and undernutrition (or, in other animals, migration 196 and hibernation), can result in chronically elevated cortisol and a shift toward greater 197 reliance on stored energy reserves. Shifts in chronic cortisol release also occur with

198 reproductive state in human females, being elevated during pregnancy to mobilize 199 maternal fat reserves to support fetal growth (Widmaier et al., 2013). 200 201 It should be noted that while cortisol promotes the release of free fatty acids via 202 lipolysis, it does not stimulate the beta oxidation pathway by which fatty acids gain entry 203 to the tricarboxylic acid cycle (Widmaier et al., 2013) Pathological elevation of cortisol 204 without increased energy expenditure, as in Cushing’s syndrome, can result in 205 excessive fat accumulation in adipose depots and ectopic locations less sensitive to 206 cortisol action and thus be associated with redistribution of fat rather than loss of fat 207 (Despres and Lemieux, 2006). 208 209 In part because cortisol is readily measured in saliva and urine, as well as in 210 blood, there are numerous field studies of human cortisol. Many studies focus on short211 term cortisol dynamics in relation to psycho-social stress. Field studies of longer term 212 effects have focused on immune function, pregnancy, and lactation (Cohen et al., 2012; 213 Janicki-Deverts et al., 2016; Nepomnaschy et al., 2006; Oaks et al., 2016; Valeggia and 214 Ellison, 2004; Valeggia and Ellison, 2003). 215 216 b. Efficiency of ATP production 217 218 At the cellular level the utilization of oxidizable substrates is auto-regulated by the 219 accumulation of downstream products, and thus ultimately driven by energy

220 expenditure. However, basal metabolic rate, the baseline turnover of energy substrates 221 in the body, is itself subject to hormonal regulation. Among the most potent regulators 222 of basal metabolism are the thyroid hormones, especially thyroxine (T4). Under 223 conditions of chronic energetic stress, such as fasting and starvation, T4 is lowered, 224 resulting in a lower baseline rate of energy consumption by the body (McAninch and 225 Bianco, 2014). 226 227 Up-regulation of T4 production can be part of an adaptive response to cold stress 228 (Leonard et al., 2005; Leppaluoto et al., 2005). The efficiency of ATP production by the 229 mitochondrial electron transport chain in some tissues, particularly brown fat, is 230 regulated by T4 through the up-regulation of uncoupling protein (UCP) (Busiello et al., 231 2015; Leppaluoto et al., 2005). UCP decouples the return flow of hydrogen ions across 232 the mitochondrial inner membrane from ATP production, resulting in an increased 233 production of heat. UCP is particularly abundant in brown adipose tissue which can 234 assist in the regulation of core body temperature, particularly in infants. Field studies of 235 high latitude populations have also demonstrated seasonal shifts in T4 production and 236 basal metabolism in adults associated with recurrent cold stress, a response that 237 appears to be greater in populations native to high latitudes than to more recent 238 migrants (Leonard et al., 2014; Levy et al., 2013; Tkachev et al., 1991). 239 240 241 Insulin-independent maintenance effort

242 243 Energy allocation is organized hierarchically. Wade and Jones (Wade and 244 Jones, 2004) schematically represent this hierarchy in three levels: functions that must 245 be maintained at or near normal levels at all times; functions that can be down246 regulated at need, but cannot easily be interrupted for long periods; and functions that 247 can be interrupted at need for extended periods. The top priority functions include the 248 maintenance of brain function via the constant maintenance of membrane 249 depolarization as well as indispensable vegetal activities such as heart and respiratory 250 function, and some aspects of kidney and liver function. Mid-level priority functions 251 include physical activity, immune function, and protein anabolism. Dispensable or 252 interruptible functions include growth and reproduction, although these are also subject 253 to mid-level down-regulation as well as interruption. 254 255 Reproductive state can reorganize this hierarchy somewhat in females. During 256 pregnancy and lactation fetal growth and infant nutrition assume relatively high 257 priorities, at or near the top level. The interruption of female fecundity by pregnancy 258 and early lactation can be viewed as evidence of the priority of the fetus and infant over 259 lower level maternal priorities. The down-regulation of maternal basal metabolism that 260 can occur during pregnancy and lactation when energy availability is limited, noted 261 above, can be seen as evidence of the higher metabolic priority assigned to the fetus 262 and infant than mid-level priorities of the mother. 263

264 Top-level metabolic priorities are largely insulin-independent (Fernandez-Real 265 and Ricart, 2003). The brain and fetus are particularly clear examples of insulin266 independent substrate uptake. Energy flow to these priorities is regulated only by the 267 availability of oxidizable substrates in the blood. The role of cortisol, glucagon, and 268 epinephrine can be best understood as buffering energy flow to top-level functions. 269 That is, when energy demands increase at lower levels, such as those precipitated by 270 “fight or flight” scenarios, these hormones increase the levels of oxidizable substrates in 271 the blood so that the new demands can be met without compromising top-level 272 functions. Down-regulation of competing demands for oxidizable substrates can also 273 increase their availability for top level functions. Physiological responses to fasting and 274 starvation, for example, include down-regulation of mid-level and low-level functions to 275 free up energy for the top priority functions (Keys et al., 1950). 276 277 Energy allocation to storage and anabolic effort 278 279 In textbooks the role of insulin at the organismic level is most often presented in 280 terms of glucose homeostasis, its function being to stimulate the clearance of excess 281 circulating glucose (Widmaier et al., 2013). But this is a poor and incomplete 282 characterization of its function. Insulin does not simply facilitate removal of glucose 283 from the blood, a function also performed by the kidney. It stimulates the uptake of 284 glucose by target tissues especially for storage in adipose tissue and to support protein 285 anabolism. Insulin also stimulates mitotic activity in many target tissues which, in

286 conjunction with its anabolic effects, can promote cellular proliferation and tissue growth 287 (Sandow, 2009). 288 289 Viewed in these terms, the key role of insulin is to promote energy allocation to 290 medium and low priority metabolic functions on a facultative basis, not simply the 291 regulation of circulating glucose levels. When metabolic energy is available in excess of 292 high and mid-level category requirements, insulin promotes the diversion of the excess 293 either to storage or to anabolism. Similarly, the “counter-regulatory hormones” cortisol, 294 glucagon, and epinephrine, do not simply counter-balance the effects of insulin on blood 295 glucose, but buffer the flow of energy to top-level metabolic functions from fluctuations 296 in intake and lower level demands independently of insulin. The balance of these 297 hormones results in the hierarchical regulation of energy flow within the body, not simply 298 in the avoidance of the pathological consequences of hyperglycemia. 299 300 Insulin was one of the first protein hormones to be isolated, characterized, and 301 used therapeutically, due to its clinical importance in the pathogenesis and treatment of 302 diabetes mellitus (Sanger and Tuppy, 1951a, b). In this context, a great deal is known 303 about short-term insulin dynamics in relation to changes in blood glucose. But recently 304 a new approach, based on measurement of C-peptide of insulin in urine, has been 305 utilized to study longer-term changes in baseline insulin levels under field conditions 306 and not in the context of disease (Sherry and Ellison, 2007). C-peptide of insulin is a 307 section of the pro-insulin molecule that is cleaved in the production of active insulin. It is

308 produced in a 1:1 ratio to active insulin and is cleared intact into the urine. Thus 309 measurement of urinary C-peptide, indexed by time, creatinine, or specific gravity, can 310 be used as a proxy measurement of insulin production. Urinary C-peptide measured in 311 samples collected in the field has been used to study longitudinal and cross-sectional 312 variation in energy balance in Samoa (Sherry and Ellison, 2007; Sherry et al., 2014), 313 Argentina (Ellison and Valeggia, 2003; Valeggia and Ellison, 2001, 2004; Valeggia and 314 Ellison, 2003), and The Gambia (Reiches et al., 2013; Reiches et al., 2014), among 315 other settings. 316 317 318 Biasing of energy allocation towards growth effort 319 320 Insulin controls energy allocation to lower priority metabolic categories, including 321 energy storage, and the anabolic categories growth and reproduction. The triage of 322 energy allocation among these competing categories, however, is largely under the 323 control of a trio of phylogenetically related protein hormones: growth hormone (GH), 324 prolactin (PRL), and human placental lactogen (hPL). The genes for growth hormone 325 and human placental lactogen are both located on chromosome 17 and display ~96% 326 sequence homology, indicating common ancestry through a gene duplication event. 327 PRL is more distantly related, located on chromosome 6 and displaying ~ 85% 328 sequence homology with GH. GH and PRL are both synthesized and secreted by 329 acidophilic cells of the anterior pituitary, while the hPL gene is expressed in the

330 placenta. All three hormones have suppressive effects on whole body insulin 331 sensitivity, primarily through an inhibition of glucose uptake by somatic adipose tissue. 332 At the same time, however, all three synergize with insulin in promoting anabolic 333 processes in target tissues: accumulation of skeletal and lean body mass in the case of 334 GH, mammary gland glucose uptake and milk synthesis in the case of PRL, and fetal 335 glucose transfer and fetal growth in the case of hPL (Forsyth and Wallis, 2002; Goffin et 336 al., 1996; Wallis et al., 2005). 337 338 The secretion of these hormones varies with maturational and reproductive 339 status, resulting in the differential allocation of available metabolic energy among these 340 anabolic categories. GH follows a steep decline following birth as the high rate of infant 341 growth declines to low levels in mid-childhood, but it undergoes an endogenous rise in 342 puberty, serving to promote skeletal and lean body growth (Bona et al., 1999; Rose et 343 al., 1991). At the same time, by increasing insulin resistance in adipose tissue it limits 344 energy allocation to storage and causes a transient rise in basal insulin levels (Guercio 345 et al., 2002). hPL by the placenta increases through pregnancy, elevating insulin 346 resistance in the mother and increasing the flow of metabolic energy to the fetus 347 (Braunstein, 2003; Mesiano and Jaffe, 2004). PRL production during lactation is 348 stimulated by infant nursing demand and acts to promote the uptake of glucose and 349 fatty acids by the mammary gland and the production of milk. At the same time it 350 increases insulin resistance in maternal adipose tissue, resulting in a diversion of 351 metabolic energy to milk production (Molitch, 2004).

352 353 There is a long history of measuring prolactin in field studies of lactation (see 354 (Ellison, 1995) and growth hormone in studies of human growth (see Bogin, 1999). 355 Studies of hPL are confined to clinical settings. 356 357 358 Energy allocation to reproductive effort 359 360 Energy allocation to reproductive effort is primarily reflected and governed by 361 gonadal steroids (Ellison, 2003b). In females, ovarian steroids directly modulate 362 fecundity, influence sexual attractiveness to males as well as receptive and proceptive 363 sexual behavior, and promote energy storage in adipose tissue (an important form of 364 somatic reproductive effort in females). In males, testosterone maintains sperm 365 production, stimulates libido and mating effort and may also support increased social 366 confidence and assertiveness, and promotes increases in muscle mass (an important 367 form of somatic reproductive effort in males). 368 369 The production and release of gonadal steroids is governed by the gonadotropin 370 hormones, follicle stimulating hormone (FSH) and luteinizing hormone (LH) secreted by 371 the anterior pituitary. However, the effectiveness of this gonadotropin stimulation 372 depends strongly on insulin, so strongly that insulin is sometimes characterized as a co373 gonadotropin (Poretsky et al., 1999; Poretsky and Kalin, 1987). This effect has been

374 mostly clearly demonstrated in in vitro studies of steroid production by cultured ovarian 375 granulose cells, where insulin receptor has been identified as the mediator of the effect, 376 increasing the rate of steroid production per cell. 377 378 Gonadal steroids, in their turn, synergize with insulin in promoting somatic 379 reproductive effort in both sexes, promoting increases in muscle mass in males 380 (androgens) and increases in adipose mass in females (estrogens) (Grumbach and 381 Styne, 2003). The mutual synergies between insulin and gonadal steroids constitute a 382 positive feedback loop that can dramatically up-regulate the flow of metabolic energy to 383 reproductive effort to take advantage of conditions of positive energy balance and the 384 availability of metabolic energy in excess of the needs of high priority categories. 385 386 The advent of practical methods for assaying gonadal steroids in saliva (Ellison, 387 1988; Lipson and Ellison, 1989), in addition to blood and urine, led to a rapid increase in 388 the number of field studies of gonadal steroids in human populations. Many examples 389 are provided elsewhere in this issue. 390 391 Hormonal gating of energy allocation 392 393 The hormonal framework described above governs the allocation of available 394 metabolic energy to competing domains in a hierarchical way. Top priority metabolic 395 categories, including brain and vegetal physiology, as well as fetal growth and infant

396 nutrition during pregnancy and lactation, receive energy in direct proportion to its 397 availability. That availability is determined by the hormones that mobilize oxidizable 398 substrates and those that govern the rate and efficiency of ATP production as well as by 399 the competing demands of physical activity. 400 401 Insulin controls the gate for allocation of energy to lower level, “dispensable” 402 categories such as energy storage, growth, and reproduction. Cortisol opposes this 403 allocation, increasing the availability of metabolic energy to mid-level categories such as 404 immune function as well as buffering top level allocations. Allocation of metabolic 405 energy among potentially competing lower level categories is governed by GH, PRL, 406 hPL, and gonadal steroids interacting with and modifying the action of insulin. 407 408 The dynamic interaction of the key hormones governing energy allocation helps 409 to organize key life history transitions. Puberty and postpartum resumption of ovarian 410 function are two particularly clear examples and will be considered here in some detail. 411 412 The pubertal transition 413 414 Puberty involves a transition in energy allocation from growth to adult 415 reproductive potential during which the body is modified, sexual dimorphism is 416 elaborated, and reproductive potential is established (Ellison et al., 2012). Although not 417 as dramatic as the metamorphosis of holometabolous insects or most amphibians, it is

418 the human equivalent, changing juvenile morphology to a distinctively adult pattern and 419 elaborating immature, nonfunctional reproductive organs into mature, functioning ones. 420 Puberty involves the close coordination of primary reproductive maturation with the 421 rapid growth and transformation of the body. Reproductive maturation involves the 422 appearance (or, more properly, reappearance) of pulsatile release of gonadotropin 423 releasing-hormone (GnRH) by the median eminence of the hypothalamus, which in turn 424 stimulates increased production of FSH and LH (Grumbach and Styne, 2003). The 425 factors that determine this change in GnRH production are still incompletely understood, 426 but the change begins quite early, well before internal or external manifestations of 427 increasing gonadal activity. The skeletal growth spurt that is typical of puberty is 428 primarily caused by an endogenous increase in GH and its downstream consequences 429 (Grumbach and Styne, 2003). Again, the causes of the increase in GH secretion are 430 incompletely understood, but its timing is roughly synchronous with the first appearance 431 of pulsatile gonadotropin secretion, suggesting a linkage between the two events (Suter, 432 2004; Gahete et al., 2016). The elaboration of somatic sexual dimorphism results from 433 the interaction of pubertal growth with rising titers of gonadal steroids. Differential 434 growth between the sexes resulting in adult sexual dimorphism is mediated by gonadal 435 steroids, both androgens and estrogens (Ellison et al., 2012). 436 437 The hormonal management of energy allocation during the pubertal transition 438 can be sketched out in terms of the framework described above. We assume that the 439 process is set in train when pulsatile GnRH reappears and GH begins to rise, even if the

440 direct causes of those two events remain to be fully understood. Rising levels of GH 441 stimulate growth and divert energy away from storage. The elevated GH also results in 442 increasing insulin resistance in adipose tissue, causing basal levels of insulin to rise. 443 Rising insulin would synergize with the increasing gonadotropin titers resulting from 444 increasingly stable GnRH pulses to promote gonadal steroid production and release. 445 As they approach mature levels, gonadal steroids in turn will potentiate the peripheral 446 effects of insulin, leading to a shift back toward lower insulin levels and increased 447 energy allocation to fat storage (in females) and muscle mass (in males), now as 448 sexually dimorphic forms of somatic reproductive effort. Even as they shift the direction 449 of energy allocation, gonadal steroids cause the closure of epiphyseal growth plates 450 and the cessation of skeletal growth as well as causing the remodeling of the female 451 pelvis. 452 453 In this way the pubertal transition unfolds as an endogenous process governed 454 by the interaction of the endocrine framework of energy allocation and its integration 455 with the growth and maturational processes involved. Energy allocation is first diverted 456 away from storage to support somatic growth and transformation and then returned to 457 storage as well as sexually dimorphic somatic forms of reproductive effort. A 458 characteristic and transient period of hyperinsulinemia is a central part of the process, 459 helping to accelerate gonadal steroid production to its mature levels. The mature 460 steroid levels in turn resolve the transient insulin resistance. 461

462 This schema is consistent with observations of shifting energy allocation priorities 463 in female Gambian adolescents (Reiches et al., 2014). In the midst of the pubertal 464 transition, when skeletal growth is still underway, Gambian girls will respond to periods 465 of energetic stress by defending lean body mass at the expense of fat mass, but later in 466 the process as skeletal growth comes to a halt the same periods of energetic stress are 467 associated with a defense of fat mass at the expense of lean body mass. The 468 metabolic priority given to growth early gives way to a metabolic priority of somatic 469 reproductive effort late (Figure 3). 470 471 A mathematical model of the pubertal transition 472 473 The smooth, autonomous nature of the transition can also be demonstrated in a 474 quantitative model of female puberty. The specifics of the model are provided in 475 Supplementary Materials. The point of the model is to demonstrate that the interactions 476 between the principal hormonal regulators reviewed above act like interlocking gears, 477 so that changes in one component drive changes in the system as a whole. The shape 478 of the changes that ensue is a function of the feedback links within the system. The 479 central set of interactions are those between insulin directing energy to anabolic effort, 480 the pituitary proteins (prolactin, growth hormone, placental lactogen) biasing energy to 481 growth effort and feeding back on insulin through their effects on somatic insulin 482 resistance, and the gonadal steroids (estradiol, progesterone, and testosterone) biasing 483 energy to reproductive effort and feeding back negatively on the pituitary proteins

484 regulating growth effort. Because the interactions of these hormones are interlocking, 485 the system as a whole can be changed by changes in any one of the components. But 486 the nature of the interactions causes the system to “switch” from a bias toward growth 487 effort to a bias toward reproductive effort, a switch that reflects the essential nature of a 488 life history transition as defined in this paper. Both the transition from growth to 489 reproduction at puberty (described here) and the resumption of ovarian function 490 postpartum (described below) can be modeled by the same system of interactions. 491 492 The assumptions and parameter settings of the quantitative models are meant to 493 be as simple as possible. In the model of the pubertal transition, GH is assumed to be 494 the independent driving factor, with all other variables dependent on it. In fact, the 495 initiation of the GH rise remains something of a mystery. There is mounting evidence 496 suggesting that the rise in GH may be tied to the same neural mechanisms, including 497 kisspeptin signaling to the pituitary, that disinhibit GnRH pulsatility in the hypothalamus 498 (Gahete et al., 2016). This would suggest that the initiation of the GH rise and the 499 nocturnal, sleep-associated appearance of pulsatile LH may be more or less coincident 500 (Apter, 1997, Suter 2004). In any event, increases in GH ordinarily precede detectable 501 increases in gonadal steroids (Rogol, 2010). So in this model the assumption is made 502 that a rise in GH starts the transition. 503 504 GH is assumed to fall steeply with age prior to puberty in parallel with growth 505 velocity (Bona et al., 1999; Rose et al., 1991; Chemaitilly et al., 2003), according to a

506 rational function (a simple mathematical expression for a pattern of smooth, asymptotic 507 decline (Otto and Day, 2007), with an endogenous, normally shaped rise and fall 508 beginning at about age 8 years and centered on age 12.5 years. These ages are 509 arbitrary, but are chosen to reflect contemporary observations of adolescent growth for 510 females in the US and other developed nations where high quality longitudinal data are 511 available (Kuczmarski et al., 2002). For males, the age parameters can be set 512 approximately two years later. The normal shape of the GH rise during puberty is an 513 arbitrary assumption that generally reflects the observed pattern (Albertsson-Wikland et 514 al., 1994). The exact shape of the trajectory isn’t important to the model; what is 515 important is that GH undergoes a rise at puberty before falling to adult levels. 516 517 Insulin, independent of the effect of GH, is assumed to follow a slow logistic rise 518 from childhood to adult levels (Ballerini et al., 2016). However, the effect of GH on 519 insulin resistance, assumed to be proportional to the level of GH, results in a transient 520 increase in insulin. Estradiol, in the absence of insulin, is assumed to follow a slow 521 logistic rise to adult levels, becoming noticeable (about 10% of adult levels) at about the 522 same time as the start of the GH rise, driven by gonadotropin levels that are themselves 523 responding to the resumption of pulsatile GnRH release. The logistic form of the 524 estradiol trajectory is arbitrary but is a simple mathematical form that can be used to 525 model sigmoid patterns (Otto and Day, 2007). A dummy factor (“lag” ) is introduced to 526 center the inflection point of the estradiol at about 12.5 years (around the time of 527 menarche) and an asymptotic approach to adult levels at around 20 years. This pattern

528 reflects empirical observations indicating continued increases in ovarian function for a 529 number of years after menarche until at least the late teens/early twenties (Lipson and 530 Ellison 1992, Ellison 1996). Under the stimulating influence of insulin, assumed to be 531 proportional to insulin level, the rise in estradiol becomes steeper and overshoots final 532 adult levels slightly before converging on them in the late teens. Growth rate is 533 assumed to be proportional to GH levels, minus a braking effect assumed to be 534 proportional to estradiol levels, resulting from the action of estradiol in accelerating the 535 closure of the epiphyses of the long bones. The breaking effect of estradiol results in a 536 more rapid deceleration of growth that reaches its steepest decline at about the time of 537 menarche. The resulting pattern is represented in Figure 4. As noted above, the 538 parameters have been arbitrarily set to reflect the timing of puberty similar to that in the 539 US and other industrialized populations. Delay in the initial rise in GH would shift the 540 entire pattern to the right, typical of later puberty. Substitution of testosterone for 541 estradiol, together with a later onset of the GH rise, would generate a model typical of 542 males. Note that testosterone conversion to estradiol in the growth plates results in the 543 same breaking effect on linear growth as in females. Note also that the model does not 544 specify exact hormone levels, but only relative levels, with 1.0 representing typical adult 545 values. 546 547 The postpartum resumption of ovarian function 548

549 A second example of the hormonal orchestration of a life history transition is the 550 postpartum resumption of ovarian function. In this case, a transition occurs between 551 energetic allocation to milk and energetic investment in fecund reproductive capacity. 552 The transition unfolds in a manner very reminiscent of the pubertal transition, with PRL 553 taking on the role played by GH during puberty. In a lactating mother early in the 554 postpartum period milk production assumes a high priority. It is driven by PRL secretion 555 which is responsive to infant demand but also reflective of maternal energy availability 556 (Ellison, 1995). When infants are exclusively breastfed by mothers facing energetic 557 constraints, PRL levels will be high and insulin levels low. PRL directs energy toward 558 milk production by increasing the insulin resistance of peripheral maternal adipose 559 tissue (though there is also evidence that PRL increases insulin sensitivity in mammary 560 adipose tissue). Gonadal activity is extremely low, though FSH levels are near normal 561 (McNeilly et al., 1994), suggesting ovarian resistance to gonadotropin stimulation. 562 563 Later in the postpartum period PRL levels begin to decline. This occurs most 564 often as a consequence of the introduction of supplementary foods into the infant’s diet 565 (McNeilly et al., 1994), reducing the demand for milk. Reduced energy demand for milk 566 production results in increased energy availability to the mother and rising levels of 567 basal insulin. Energy allocation to storage rises as a consequence, but PRL levels 568 remain sufficient to cause elevated insulin resistance. As a result, insulin levels rise 569 above the typical level for the mother, manifesting a brief, transient period of 570 hyperinsulinemia. This transient period of hyperinsulinemia, although briefer in

571 duration, is very reminiscent of the transient period of elevated insulin in puberty. The 572 elevated insulin synergize with FSH levels to stimulate ovarian steroid production 573 toward normal mature levels. The rising titers of estradiol that result in turn potentiate 574 energy storage in adipose tissue, increasing adipose insulin sensitivity and returning 575 insulin levels to normal. 576 577 As with the pubertal transition, the resumption of postpartum ovarian function is 578 governed endogenously by the endocrine architecture of energy allocation. Insulin once 579 again plays a central role, modified by the actions of PRL and gonadal steroids. The 580 unfolding sequence is clearly displayed by Toba mothers in Argentina (Ellison and 581 Valeggia, 2003; Valeggia and Ellison, 2001, 2004; Valeggia and Ellison, 2003) (Figure 582 5), and a comparable pattern has even been observed in wild chimpanzees in Uganda 583 (Emery Thompson et al., 2012). 584 585 A mathematical model of the postpartum resumption of ovarian function 586 587 As with the pubertal transition, the smooth, endogenous nature of the resumption 588 of ovarian function postpartum can be represented in a qualitative model (details in 589 Supplementary Materials). This model includes analogous hormonal interactions to 590 those presented in the puberty model above. In this case, PRL level is assumed to be 591 the independent driving factor with other variables being dependent on it. Prolactin is 592 assumed to decline following a logistic function (a simple expression for a sigmoid

593 pattern, Otto and Day, 2007) and to represent the energy demand for milk production. 594 In the version of the model presented here the decline in PRL is parameterized to return 595 to baseline by 40 months with a maximal rate of decline at 18-20 months. This is an 596 arbitrary parameterization designed to reflect patterns observed in the Toba (Valeggia 597 and Ellison, 2004). Different values for the parameters of the logistic could be chosen 598 to reflect earlier or later weaning or other factors affecting energy availability. Energy 599 availability for lower level priorities than milk production is assumed to be proportional to 600 the complement of PRL (where peak PRL is set at 1.0 as a reference). Insulin is 601 expected to be proportional to energy availability, but modified by the insulin resistance 602 caused by prolactin. If insulin sensitivity were constant, insulin levels would rise 603 proportionally to energy availability. However, due to the declining effect of prolactin on 604 insulin resistance, insulin rises more steeply and overshoots typical “post lactation” 605 levels (set at 1.0). Estradiol levels rise proportionally to insulin with a slight, arbitrary lag 606 time, while postpartum weight gain is proportional to the product of energy availability 607 and insulin. The resulting pattern is represented in Figure 6 with parameters adjusted to 608 be roughly equivalent to observations made on the Toba. Once again, variable values 609 are relative with 1.0 representing adult, non-lactating values of all variables except PRL, 610 where 1.0 represents peak values early in lactation. 611 612 The heuristic value of the mathematical model is to make explicit the fact that the 613 mutual interactions of a core set of hormones regulating energy allocation, interactions 614 that are well-established in the literature, are sufficient to generate the rather complex

615 trajectories of hormones and anabolic variables (e.g., growth rate, milk production, 616 weight gain) that are observed during two important life history transitions, puberty and 617 the post-partum resumption of ovarian function. It is not necessary to assume any 618 special set of interactions or drivers to generate these patterns. Rather the “switch” 619 from growth effort to reproductive effort is latently embedded in the effect these 620 important regulators have on each other. Natural selection has been able to leverage 621 the same endocrine architecture to orchestrate two different life history transitions by 622 making use of two different, but phylogenetically related, modifiers of anabolic energy 623 allocation, GH and PRL. Although beyond the scope of this paper, it can be argued that 624 this architecture has analogues in other vertebrates (Chandrashekar and Bartke, 2003, 625 Kawashima et al., 2007, Flatt and Heyland, 2011). 626 627 628 Concluding comments 629 630 The model framework presented here is just that: a model that represents only a 631 few of the major features of a complex network or interacting signals that govern human 632 energetic allocation. Its heuristic value is illustrated in its ability to capture the major 633 signals and interactions in sufficient detail to illuminate the dynamic aspects of the 634 control of energy allocation during major life history transitions. In doing so, it helps to 635 connect our knowledge of the mechanisms that govern energy allocation with the 636 predictions and tradeoffs that feature in life history theory. As a major branch of

637 evolutionary theory, life history theory has proven very powerful in predicting and 638 explaining major features of life history diversity, and in doing so it leans very heavily on 639 generalized concepts of energetic tradeoffs. But less effort has been made to integrate 640 life history theory with proximate mechanisms until recently (Flatt and Heyland, 2011). 641 642 The model presented here also underscores the central role of insulin, not as a 643 gluco-regulatory hormone, but as the major gatekeeper of energy allocation to mid- and 644 lower level physiological priorities. Insulin does not simple clear glucose from the 645 circulation to guard against negative effects of hyperglycemia, it directs energy toward 646 growth and reproductive effort, synchronizing investment in those physiological 647 categories with the availability of metabolic energy over and above the requirements of 648 higher priority categories. This centrality of insulin in the modulation of energy 649 allocation helps to make sense of the phylogenetically highly conserved relationship 650 between insulin and insulin-like signaling and lifespan variation (Barbieri et al., 2003; 651 Singleton, 2011; Tatar et al., 2003). Insulin is one of the best known and longest 652 studied human hormones. It may not have the cachet of more recently identified 653 neuropeptides and cytokines, nor does it represent control of the soma by the CNS. Yet 654 its role in life history energetics is crucial. 655 656 Finally, although developed in the context of human physiology, there is reason 657 to suspect that the framework presented here may have more general application, 658 either directly or as a template to modify and build on. The comparability of the

659 trajectory of C-peptide of insulin in relation to the postpartum resumption of ovarian 660 function in humans and chimpanzees is one example that supports this notion. The 661 framework developed and presented here may serve as an impetus for comparative 662 studies of the hormonal architecture of energy allocation in non-human primates and 663 mammals generally. 664

665 References 666 667 Albertsson-Wikland, K., Rosberg, S., Karlberg, J., Groth, T., 1994. Analysis of 24-hour 668 growth hormone profiles in healthy boys and girls of normal stature: relation to puberty. 669 J Clin Endocrinol Metab 78(5): 1195-1201. 670 Apter, D., 1997. Development of the hypothalamic-pituitary-ovarian axis. Ann N Y 671 Acad Sci 816:9-21. 672 Badman, M.K., Flier, J.S., 2005. The gut and energy balance: visceral allies in the 673 obesity wars. Science 307, 1909-1914. 674 Ballerini, M.G., Bergadá, I., Rodríguez, M.E., Keselman, A., Bengolea, V.S., Pipman, V., 675 Domené, H.M., Jasper, H.G., Ropelato, M.G., 2016. Arch Argent Pediatr 114(4):328676 336. 677 Barbieri, M., Bonafe, M., Franceschi, C., Paolisso, G., 2003. Insulin/IGF-I-signaling 678 pathway: an evolutionarily conserved mechanism of longevity from yeast to humans. 679 Am J Physiol Endocrinol Metab 285, E1064-1071. 680 Blundell, J.E., Finlayson, G., Gibbons, C., Caudwell, P., Hopkins, M., 2015a. The 681 biology of appetite control: Do resting metabolic rate and fat-free mass drive energy 682 intake? Physiol Behav 152, 473-478. 683 Blundell, J.E., Gibbons, C., Caudwell, P., Finlayson, G., Hopkins, M., 2015b. Appetite 684 control and energy balance: impact of exercise. Obes Rev 16 Suppl 1, 67-76. 685 Bogin, B.A., 1999. Patterns of Human Growth, 2nd Ed., 2 ed. Cambridge University 686 Press, Cambridge, UK.

687 Bona, G., Petri, A., Conti, A., Sartorio, A., 1999. The impact of gender, puberty and 688 body mass on reference values for urinary growth hormone (GH) excretion in normally 689 growing non-obese and obese children. Clinical Endocrinology 50, 775-781. 690 Braunstein, G.D., 2003. Endocrine changes in pregnancy, in: Larsen, P.R., Kronenberg, 691 H.M., Melmed, S., Polonsky, K.S. (Eds.), Williams textbook of endocrinology, 10th Ed., 692 10 ed. Saunders, Philadelphia, PA, pp. 795-810. 693 Bribiescas, R.G., 2001. Serum leptin levels and anthropometric correlates in Ache 694 Amerindians of eastern Paraguay. Am J Phys Anthropol 115, 297-303. 695 Bribiescas, R.G., 2005. Serum leptin levels in Ache Amerindian females with normal 696 adiposity are not significantly different from American anorexia nervosa patients. 697 American journal of human biology : the official journal of the Human Biology Council 698 17, 207-210. 699 Bribiescas, R.G., Hickey, M.S., 2006. Population variation and differences in serum 700 leptin independent of adiposity: a comparison of Ache Amerindian men of Paraguay and 701 lean American male distance runners. Nutr Metab 3, 34-40. 702 Busiello, R.A., Savarese, S., Lombardi, A., 2015. Mitochondrial uncoupling proteins and 703 energy metabolism. Front Physiol 6, 36. 704 Chemaitilly, W., Trivin, C., Souberbielle, J. C., Brauner, R., 2003. Assessing short705 statured children for growth hormone deficiency. Horm Res 60(1):34-42. 706 Cohen, S., Janicki-Deverts, D., Doyle, W.J., Miller, G.E., Frank, E., Rabin, B.S., Turner, 707 R.B., 2012. Chronic stress, glucocorticoid receptor resistance, inflammation, and

708 disease risk. Proceedings of the National Academy of Sciences of the United States of 709 America 109, 5995-5999. 710 Crespo, C.S., Cachero, A.P., Jiménez, L.P., Barrios, V., Ferreiro, E.A., 2014. Peptides 711 and food intake. Frontiers in Endocrinology 5. 712 Despres, J.P., Lemieux, I., 2006. Abdominal obesity and metabolic syndrome. Nature 713 444, 881-887. 714 Ellison, P.T., 1988. Human salivary steroids: methodological considerations and 715 applications in physical anthropology. Yearbook of Physical Anthropology 31, 115-132. 716 Ellison, P.T., 1995. Breasfeeding, fertility, and maternal condition, in: Dettwyler, K.A., 717 Stuart-Macadam, P. (Eds.), Breasfeeding: Biocultural perspectives. Aldine de Gruyter, 718 Hawthorne, NY, pp. 305-345. 719 Ellison, P.T., 1996. Developmental influences on adult ovarian function. Am J Hum 720 Biol 8:725-734. 721 Ellison, P.T., 2003a. Energetics and reproductive effort. American Journal of Human 722 Biology 15, 342-351. 723 Ellison, P.T., 2003b. Energetics and reproductive effort. American journal of human 724 biology : the official journal of the Human Biology Council 15, 342-351. 725 Ellison, P.T., 2009. Social relationships and reproductive ecology, in: Ellison, P.T., Gray, 726 P. (Eds.), Endocrinology of social relationships. Harvard University Press, Cambridge, 727 MA, pp. 54-73. 728 Ellison, P.T., Reiches, M.W., Shattuck-Faegre, H., Breakey, A., Konecna, M., Urlacher, 729 S., Wobber, V., 2012. Puberty as a life history transition. Ann Hum Biol 39, 352-360.

730 Ellison, P.T., Valeggia, C.R., 2003. C-peptide levels and the duration of lactational 731 amenorrhea. Fertility and sterility 80, 1279-1280. 732 Emery Thompson, M., Muller, M.N., Wrangham, R.W., 2012. The energetics of lactation 733 and the return to fecundity in wild chimpanzees. Behavioral Ecology 23, 1234-1241. 734 Fernandez-Real, J.M., Ricart, W., 2003. Insulin resistance and chronic cardiovascular 735 inflammatory syndrome. Endocrine reviews 24, 278-301. 736 Flatt, T., Heyland, A., 2011. Mechanisms of life history evolution: The genetics and 737 physiology of life history traits and trade-offs. Oxford University Press, Oxford. 738 Forsyth, I.A., Wallis, M., 2002. Growth hormone and prolactin--molecular and functional 739 evolution. Journal of mammary gland biology and neoplasia 7, 291-312. 740 Gahete, M.D., Vázquez-Borrego, M.C., Martínez-Fuentes, A.J., Tena-Sempere, M., 741 Castaño, J.P., Luque, R.M., 2016. Role of the Kiss1/Kiss1r system in the regulation of 742 pituitary cell function. Mol Cell Endocrinol epub ahead of print (doi: 743 10.1016/j.mce.2016.07.039).
744 745 Goffin, V., Shiverick, K.T., Kelly, P.A., Martial, J.A., 1996. Sequence-function 746 relationships within the expanding family of prolactin, growth hormone, placental 747 lactogen, and related proteins in mammals. Endocrine reviews 17, 385-410. 748 Grumbach, M.M., Styne, D.M., 2003. Puberty: ontogeny, neuroendocrinology, 749 physiology, and disorders, in: Larsen, P.R., Kronenberg, H.M., Melmed, S., Polonsky, 750 K.S. (Eds.), Williams textbook of endocrinology, 10th Ed., 10 ed. Saunders, 751 Philadelphia, pp. 1115-1286.

752 Guercio, G., Rivarola, M.A., Chaler, E., Maceiras, M., Belgorosky, A., 2002. 753 Relationship between the GH/IGF-I axis, insulin sensitivity, and adrenal androgens in 754 normal prepubertal and pubertal boys. The Journal of clinical endocrinology and 755 metabolism 87, 1162-1169. 756 Janicki-Deverts, D., Cohen, S., Turner, R.B., Doyle, W.J., 2016. Basal salivary cortisol 757 secretion and susceptibility to upper respiratory infection. Brain Behav Immun. epub 758 ahead of print (http://dx.doi.org/10.1016/j.bbi.2016.01.013 )
759 Kawashima, C., Fukihara, S., Maeda, M., Kaneko, E., Montoya C.A., Matsui, M., 760 Shimizu, T., Matsunaga, N., Kida, K., Miyake, Y.-I., Schams, D., Miyamoto, A., 2007. 761 Reproduction 133:155-163. 762 Keys, A., Brozek, J., Henschel, A., Mickelsen, O., Taylor, H.L., 1950. The biology of 763 human starvatin, vols. I-II. University of Minnesota Press, Minneapolis, MN. 764 Kuczmarski RJ, Ogden CL, Guo SS, et al. 2000. CDC growth charts for the United 765 States: Methods and development. National Center for Health Statistics. Vital Health 766 Stat 11(246). 2002
767 Kuzawa, C.W., Quinn, E.A., Adair, L.S., 2007. Leptin in a lean population of Filipino 768 adolescents. Am J Phys Anthropol 132, 642-649. 769 Leonard, W.R., Levy, S.B., Tarskaia, L.A., Klimova, T.M., Fedorova, V.I., Baltakhinova, 770 M.E., Krivoshapkin, V.G., Snodgrass, J.J., 2014. Seasonal variation in basal metabolic 771 rates among the Yakut (Sakha) of Northeastern Siberia. American journal of human 772 biology : the official journal of the Human Biology Council 26, 437-445.

773 Leonard, W.R., Snodgrass, J.J., Sorensen, M.V., 2005. Metabolic adaptation in 774 indigenous Siberian populations. Annual Review of Anthropology 34, 451-471. 775 Leppaluoto, J., Paakkonen, T., Korhonen, I., Hassi, J., 2005. Pituitary and autonomic 776 responses to cold exposures in man. Acta Physiol Scand 184, 255-264. 777 Levy, S.B., Leonard, W.R., Tarskaia, L.A., Klimova, T.M., Fedorova, V.I., Baltakhinova, 778 M.E., Krivoshapkin, V.G., Snodgrass, J.J., 2013. Seasonal and socioeconomic 779 influences on thyroid function among the Yakut (Sakha) of Eastern Siberia. American 780 journal of human biology : the official journal of the Human Biology Council 25, 814-820. 781 Lipson, S.F., Ellison, P.T., 1989. Development of protocols for the applicatin of salivary 782 steroid analyses to field conditions. American Journal of Human Biology 1, 249-255. 783 Lipson, S.F., Ellison, P.T., 1992. Normative study of age variation in salivary 784 progesterone profiles. Journal of Biosocial Science 24:233-244. 785 McAninch, E.A., Bianco, A.C., 2014. Thyroid hormone signaling in energy homeostasis 786 and energy metabolism. Annals of the New York Academy of Sciences 1311, 77-87. 787 McNeilly, A.S., Tay, C.C., Glasier, A., 1994. Physiological mechanisms undelying 788 lactational amenorrhea. Ann N Y Acad Sci 709:145-155. 789 Meier, U., Gressner, A.M., 2004. Endocrine regulation of energy metabolism: review of 790 pathobiochemical and clinical chemical aspects of leptin, ghrelin, adiponectin, and 791 resistin. Clin Chem 50, 1511-1525. 792 Mesiano, S., Jaffe, R.B., 2004. The endocrinology of human pregnancy an feto793 placental neuroendocrine development, in: Straus, J.F., III, Barbieri, R.L. (Eds.), Yen

794 and Jaffe's Reproductive Endocrinology, 5th Ed., 5 ed. Elsevier Saunders, Philadelphia, 795 PA, pp. 327-367. 796 Miller, A.A., Sharrock, K.C., McDade, T.W., 2006. Measurement of leptin in dried blood 797 spot samples. American journal of human biology : the official journal of the Human 798 Biology Council 18, 857-860. 799 Molitch, M.E., 2004. Prolactin in human reproduction, in: Straus, J.F., III, Barbieri, R.L. 800 (Eds.), Yen and Jaffe's Reproductive Endocrinology, 5 ed. Elsevier Saunders, 801 Philadelphia, PA, pp. 93-124. 802 Munch-Andersen, T., DSorensen, K., Aachmann-Andersen, N.-J., Aksglaede, L., Juul, 803 A., Helge, J.W., 2013. Ethnic differences in leptin and adiponectin levels between 804 Greenlandic Inuit and Danish children. Int J Circumpolar Health 72, 21458. 805 Nepomnaschy, P.A., Welch, K.B., McConnell, D.S., Low, B.S., Strassmann, B.I., 806 England, B.G., 2006. Cortisol levels and very early pregnancy loss in humans. 807 Proceedings of the National Academy of Sciences of the United States of America 103, 808 3938-3942. 809 Oaks, B.M., Laugero, K.D., Stewart, C.P., Adu-Afarwuah, S., Lartey, A., Ashorn, P., 810 Vosti, S.A., Dewey, K.G., 2016. Late-pregnancy salivary cortisol concentrations of 811 Ghanaian women participating in a randomized controlled trial of prenatal lipid-based 812 nutrient supplements. J Nutr. 813 Otto, S. P., Day, T, 2007. A Biologist's Guide to Mathematical Modeling in Ecology and 814 Evolution. Princeton University Press, Princeton, NJ.

815 Pelleymounter, M.A., Cullen, M.J., Baker, M.B., Hecht, R., Winters, D., Boone, T., 816 Collins, F., 1995. Effects of the obese gene product on body weight regulation in ob/ob 817 mice. Science 269, 540-543. 818 Pinkney, J., 2014. The role of ghrelin in metabolic regulation. Current opinion in clinical 819 nutrition and metabolic care 17, 497-502. 820 Poretsky, L., Cataldo, N.A., Rosenwaks, Z., Giudice, L.C., 1999. The insulin-related 821 ovarian regulatory system in health and disease. Endocrine reviews 20, 535-582. 822 Poretsky, L., Kalin, M.F., 1987. The gonadotropic function of insulin. Endocrine reviews 823 8, 132-141. 824 Reiches, M.W., E., M.S., Prentice, A.M., Prentice, A., Sawo, Y., Ellison, P.T., 2013. The 825 adolescent transition under stress: body composition tradeoffs among adolescent 826 women in The Gambia. Evol Med Publ Health 2013, 75-85. 827 Reiches, M.W., Moore, S.E., Prentice, A.M., Ellison, P.T., 2014. Endocrine responses, 828 weight change, and energy sparing mechanisms during Ramadan among Gambian 829 adolescent women. American journal of human biology : the official journal of the 830 Human Biology Council 26, 395-400. 831 Rogol, A.D., 2010. Sex steroids, groeth hormone, leptin, and the pubertal growth spurt. 832 Endocr Dev 17:77-85. 833 Rose, S.R., Municchi, G., Barnes, K.M., Kamp, G.A., Uriarte, M.M., Ross, J.L., 834 Cassorla, F., Cutler, G.B., Jr., 1991. Spontaneous growth hormone secretion increases 835 during puberty in normal girls and boys. The Journal of clinical endocrinology and 836 metabolism 73, 428-435.

837 Sandow, J., 2009. Growth effects of insulin and insulin analogues. Archives of 838 physiology and biochemistry 115, 72-85. 839 Sanger, F., Tuppy, H., 1951a. The amino-acid sequence in the phenylalanyl chain of 840 insulin. 2. The investigation of peptides from enzymic hydrolysates. The Biochemical 841 journal 49, 481-490. 842 Sanger, F., Tuppy, H., 1951b. The amino-acid sequence in the phenylalanyl chain of 843 insulin. I. The identification of lower peptides from partial hydrolysates. The Biochemical 844 journal 49, 463-481. 845 Schwartz, G.J., 2000. The role of gastrointestinal vagal afferents in the control of food 846 intake: current prospects. Nutrition 16, 866-873. 847 Schwartz, M.W., Woods, S.C., Porte, D., Jr., Seeley, R.J., Baskin, D.G., 2000. Central 848 nervous system control of food intake. Nature 404, 661-671. 849 Sharrock, K.C., Kuzawa, C.W., Leonard, W.R., Tanner, S., Reyes-Garcia, V.E., Vadez, 850 V., Huanca, T., McDade, T.W., 2008. Developmental changes in the relationship 851 between leptin and adiposity among Tsimane children and adolescents. American 852 journal of human biology : the official journal of the Human Biology Council 20, 392-398. 853 Sherry, D.S., Ellison, P.T., 2007. Potential applications of urinary C-peptide of insulin for 854 comparative energetics research. Am J Phys Anthropol 133, 771-778. 855 Sherry, D.S., McGarvey, S.T., Sesepasara, M.L., Ellison, P.T., 2014. Ovarian function in 856 Samoan women shows stronger association with signals of energy metabolism than fat 857 reserves. American journal of human biology : the official journal of the Human Biology 858 Council 26, 95-98.

859 Singleton, A.W., 2011. Evolution and the regulation of growth and body size, in: Flatt, 860 T., Heyland, A. (Eds.), Mechanisms of life history evolution: The genetics and 861 physiology of life history traits and tradeoffs. Oxford University Press, New York, pp. 43862 55. 863 Suter, K.J., 2004. The ontogeny of pulsatile growth hormone secretion and its temporal 864 relationship to the onset of puberty in the agonadal male rhesus monkey (Macaca 865 mulatta). J Clin Endocrinol Metab 89(5):2275-2280. 866 Tanaka, M., Umezaki, M., Natsuhara, K., Yamauchi, T., Inaoka, T., Hongo, T., Nagano, 867 M., Watanabe, C., Ohtsuka, R., 2005. No difference in serum leptin concentrations 868 between urban-dwelling Austronesians and Non-Austronesians in Papua New Guinea. 869 American journal of human biology : the official journal of the Human Biology Council 870 17, 696-703. 871 Tatar, M., Bartke, A., Antebi, A., 2003. The endocrine regulation of aging by insulin-like 872 signals. Science 299, 1346-1351. 873 Tkachev, A.V., Bojko, J.R., Ramenskaya, E.B., 1991. Endocrine status and plama lipids 874 in inhabitants of the norther European part of the USSR. Arctic Medical Research 50, 875 148-151. 876 Valeggia, C., Ellison, P.T., 2001. Lactation, energetics, and postpartum fecundity, in: 877 Ellison, P.T. (Ed.), Reproductive ecology and human evolution. Aldine de Gruyter, New 878 York, pp. 85-105. 879 Valeggia, C., Ellison, P.T., 2004. Lactational amenorrhoea in well-nourished Toba 880 women of Formosa, Argentina. J Biosoc Sci 36, 573-595.

881 Valeggia, C.R., Ellison, P.T., 2003. Impact of breastfeeding on anthropometric changes 882 in peri-urban Toba women (Argentina). American journal of human biology : the official 883 journal of the Human Biology Council 15, 717-724. 884 Wade, G.N., Jones, J.E., 2004. Neuroendocrinology of nutritional infertility. Am J 885 Physiol Regul Integr Comp Physiol 287, R1277-1296. 886 Wallis, O.C., Mac-Kwashie, O., Makri, G., Wallis, M., 2005. Molecular evolution of 887 prolactin in primates. Journal of Molecular Evolution 60, 606-614. 888 Webber, E.S., Bonci, A., Krashes, M.J., 2015. The elegance of energy balance: Insight 889 from circuit-level manipulations. Synapse 69, 461-474. 890 Widmaier, E., Raff, H., Strang, K., 2013. Vander's Human Physiology, 13th Ed. 891 McGraw-Hill, New York. 892 Wolf, G., 1996. Leptin: the weight-reducing plasma protein encoded by the obese gene. 893 Nutr Rev 54, 91-93. 894 895

896 Figure Captions 897 898 899 Figure 1: The basic framework of energy flow underlying human life history energetics. 900 Numbers associated with the arrows refer to the groups of hormonal regulators 901 specified in Figure 2. The parenthetical number associated with energy flow from 902 available metabolic energy to activity indicates potentially weak or indirect hormonal 903 regulation. 904 905 906 Figure 2: Categories of hormonal regulators associated with the pathways of energy 907 flow specified in Figure 1. Examples of major hormonal regulators in each category are 908 given and further elaborated upon in the text. 909 910 911 Figure 3: Body composition changes (means ± SE) in adolescent females in The 912 Gambia during periods of relative energy abundance (harvest season) and energy 913 constraint (hungry season) subdivided by growth rate. Details provided in Reiches et al. 914 2014. 915 916

917 Figure 4: Trajectories of growth hormone, growth rate, estradiol, and insulin during 918 puberty generated by a simple model of the basic framework of human life history 919 energetics presented in this paper and illustrated in Figure 1. The Y-axis units are 920 expressed in proportion to adult values for each variable. Details of the model are 921 provided in Supplementary Materials. 922 923 924 Figure 5: Plots of first morning urinary C-peptide of insulin levels expressed as a 925 proportion of individual, post-resumption average levels for each individual during the 926 postpartum lactation period as observed among Toba women compared with the 927 trajectories of (A) BMI and (B) urinary estrone conjugates (urinary metabolite of 928 estradiol). Compare these trajectories with those generated by the model depicted in 929 Figure 5. Data from Ellison and Valeggia 2003, Valeggia and Ellison 2001, 2003, 2004. 930 931 932 Figure 6: Trajectories of insulin, estradiol, postpartum weight gain, energy availability, 933 and prolactin during the postpartum lactating period generated by a simple model of the 934 basic framework of human life history energetics presented in this paper and illustrated 935 in Figure 1. The Y-axis units are expressed in proportion to adult values for each 936 variable. Details of the model are provided in Supplementary Materials. 937 938

939