Parental provisioning of offspring is an intensive energetic investment that is expected to compromise future offspring production. This trade-off is particularly salient for mammals in which mothers bear the exclusive burden of lactation and draw from their own energy reserves to provision offspring. The degree to which lactation impacts future reproductive ability should vary not only by the absolute cost of milk production but also by the ability of individual mothers to afford it. Few studies have been able to quantify the costs of lactation or how they affect reproductive rates. Here, we examine the metabolic load of lactation in chimpanzees, a species with intensive parental investment and extremely slow reproductive rates. We used urinary C-peptide of insulin to trace changes in energetic condition in 17 wild, unprovisioned chimpanzee mothers in Kibale National Park, Uganda. C-peptide levels of nursing mothers were depressed for six months postpartum, thereafter showing a net increase through the second year. These changes are consistent with milestones of infant physical and behavioral development. Mothers inhabiting lower quality foraging areas experienced a higher metabolic load of lactation, as indicated by lower C-peptide profiles than mothers in food-rich areas. Ovarian activity was closely correlated with energetic condition. Cycling resumed only after a sustained period of energy gain, suggesting that the slow reproductive pattern in wild chimpanzees results not only from the direct expense of milk production but also from the long period that mothers require to recover their physical condition in a food-limited environment.
Nutritional provisioning of offspring is an integral part of reproduction. Although the extent and form of provisioning varies dramatically, provisioning investment has positive effects on indices of offspring quality, such as body size and survival to maturity, in a broad range of taxa (reptiles: Sinervo 1990; mammals: Fedak et al. 1996; fish: Einum and Fleming 1999; insects: Hunt and Simmons 2000; birds: Schwagmeyer and Mock 2008; amphibians: Ficetola et al. 2011). Such studies emphasize that resource availability constrains the ability of parents to invest to the degree that would maximize offspring quality. Instead, provisioning strategies are expected to balance the advantages of parental investment for offspring quality with the associated costs to parental survival and reproductive output (Roff 1992; Stearns 1992). This trade-off is particularly salient for females who bear a disproportionate burden of parental investment in most species (Trivers 1972).
The sex difference in parental investment is particularly pronounced among mammals for which internal gestation and lactation are exclusively maternal obligations. Although lactation has the advantage of allowing mothers to draw from their own energy reserves to provision offspring when resources are scarce in the environment, it is a relatively inefficient process for energy transfer that increases the costs associated with reproduction along with the risks to maternal health (Dall and Boyd 2004). Lactation is also the most energetically demanding part of reproduction for a female mammal (Hanwell and Peaker 1977; Gittleman and Thompson 1988; Clutton-Brock et al. 1989). The total costs of milk production are expected to be particularly high for species, such as many primates, that have relatively large neonate sizes and high investment in brain tissue (Gittleman and Thompson 1988; Martin 1995). Infant feeding costs are distributed over a substantially longer period in primates than in other mammals of similar size, with dilute milks and slow infant growth rates reducing daily costs of lactation (Martin 1984; Oftedal 1984; Dufour and Sauther 2002). However, this adaptation results in long intervals between births, meaning that the cumulative energetic costs of lactation remain a significant constraint on reproductive rates (Lee 1996). Consequently, energy management during lactation is a vital influence on reproductive success via both infant health and length of the interbirth interval.
Numerous observational studies of primates and other mammals have documented behavioral and metabolic strategies to increase energy availability during lactation. Mothers may increase foraging effort, focus on higher quality food items, spend less time in energetically costly behaviors, burn fat reserves, or alter metabolic processes to subsidize the costs of milk production (Pond 1977; Gittleman and Thompson 1988; Dufour and Sauther 2002). Several species show evidence of higher infant survivorship or shorter interbirth intervals when mothers are in better energetic condition (Papio anubis: Bercovitch 1987, Rosetta et al. 2011; Propithecus verreauxi: Richard et al. 2000; Phoca vitulina: Bowen et al. 2001; Callithrix jacchus: Tardif et al. 2001; Pan troglodytes: Emery Thompson, Kahlenberg et al. 2007; Odocoileus virginianus: Therrien et al. 2008). More recent studies suggest that the nutritional quality of breast milk, which varies with energetic condition of the mother, can also influence infant behavioral development (Macaca mulatta: Hinde and Capitanio 2010).
Despite the apparent importance of energy management during lactation for female reproductive success, relatively little is known about how the costs of milk production are distributed over time and how variation in energy availability impacts lactation and reproductive rates. Such data are particularly hard to acquire from wild primates in energy-limiting environments. Observational studies of suckling are problematic because it is difficult to quantify actual milk transfer versus transient nipple contact and because a considerable amount of nursing occurs out of view of observers (e.g., at night). It is challenging to safely obtain weights from many wild primates, particularly at intervals necessary to monitor short-term changes in condition, and mothers and their clinging infants can usually not be weighed independently.
One feasible approach to measure the metabolic costs of lactation is to monitor the production of insulin, which can be assessed noninvasively through urinary C-peptide (Rubenstein et al. 1969; Melani et al. 1970). Insulin coordinates the storage and uptake of energy in somatic tissues and serves as a specific signal and regulator of long-term energy balance in the brain (Schwartz et al. 1992; Strack et al. 1995; Norman and Litwack 1997; Havel 2001). In addition to being a general indicator of energy balance, insulin plays a direct role in reproduction, promoting ovarian steroid production (Garzo and Dorrington 1984; Willis et al. 1996; Greisen et al. 2001). During pregnancy and lactation, changing insulin dynamics are fundamental to managing maternal energy stores and increasing energy available to the newborn (Durnin 1987; Catalano et al. 1991; Catalano et al. 1993; Butte and Hopkinson 1998; Catalano et al. 1998; Tigas et al. 2002). Early lactation in human and animal models is characterized by relatively low insulin levels, leading to greater bioavailability of glucose for lactogenesis and changes in insulin sensitivity that favor energy uptake by mammary tissue and reduce use by peripheral tissues (Flint et al. 1979; Jones et al. 1984; Vernon 1989; Tigas et al. 2002). Thus, insulin or C-peptide analyses may help reveal variability in the metabolic costs of lactation. For example, experimentally malnourished rats exhibited decreased plasma insulin levels during lactation compared with controls (Holemans et al. 1996). Similarly, relatively lean human mothers produced less insulin, measured via urinary C-peptide, during early lactation and experienced a shallower increase in insulin production over time (Ellison and Valeggia 2003). Ellison and Valeggia (2003) (Valeggia and Ellison, 2004, 2009) also linked increases in individually normalized C-peptide levels with postpartum cycle resumption in lactating Toba women. C-peptide trajectories mirrored changes in maternal BMI, and both elevated C-peptide levels and a positive change in BMI occurred consistently prior to the return of menses.
Here, we examine the costs of maternal provisioning in a species with one of the slowest reproductive rates of any animal. Wild chimpanzees (Pan troglodytes) produce infants approximately once every 5–7 years (Emery Thompson, Jones, et al. 2007), and infant provisioning occurs primarily, though not exclusively, via breast milk (Goodall 1986; Hiraiwa-Hasegawa 1990). We monitored C-peptide profiles over the course of lactational amenorrhea in order to test the hypothesis that female reproductive rates are limited by variation in the metabolic load of lactation. This study addresses 3 related questions. First, how are the energetic costs of milk production distributed over the course of lactation? Observational data suggest that chimpanzee infants are not fully weaned until 4–5 years of age and indicate supplementation with solid foods and a decline in nursing frequency within the first year (Clark 1977; Rijt-Plooij and Plooij 1987). Therefore, by describing the average trajectories of C-peptide change during lactation, we sought to determine how early changes in lactation intensity, versus the cessation of nursing, affect female energetic condition and how the duration of intense energy expenditure for milk production relates to the average length of lactational amenorrhea in this species. This first objective focuses on general patterns. Our second question addresses variation within these patterns: Do C-peptide levels predict reproductive function? To address this question, we first asked whether C-peptide levels predict the production of ovarian steroid hormones in individuals. Next, we asked whether the resumption of cycling was temporally associated with an increase in C-peptide levels, following the methods of Ellison and Valeggia (2003). Finally, does maternal energy availability affect the metabolic load of lactation? To examine this question, we compared the C-peptide profiles of 2 clusters of mothers who experience a systematic difference in resource access due to the location of their core foraging areas.
MATERIALS AND METHODS
Study site and subjects
Data collection focused on the Kanyawara community of wild, unprovisioned chimpanzees living in the northwestern portion of Kibale National Park, Uganda. The Kanyawara chimpanzees occupy a relatively large home range (14–30 km2, Wilson et al. 2012), consisting primarily of semi-deciduous forest interspersed with grasslands, swamp, and colonizing forests (Chapman and Wrangham 1993). Compared with the neighboring Ngogo site, Kanyawara has lower densities of chimpanzee foods, resulting in decreased diet quality and energy balance for its residents (Chapman et al. 1997; Emery Thompson et al. 2009; Potts et al. 2011). This may account for the relatively slow birth rates (>6.5 year interval to next birth following surviving infant) at Kanyawara compared with other long-term chimpanzee study sites (Emery Thompson et al. 2006; Emery Thompson, Jones, et al. 2007).
In order to test the prediction that individual differences in resource access affect the metabolic load of lactation, we assigned females to 1 of 2 categories, according to their use of habitat, following previous quantitative analysis of ranging, diet, and association patterns in this community (Emery Thompson and Wrangham 2006; Emery Thompson, Kahlenberg, et al. 2007). Central females primarily utilized the central and southern portions of the home range, which have rich concentrations of preferred fruit trees, while peripheral females focused their activities to the north, where fruit is less abundant (Skorupa 1988; Emery Thompson, Kahlenberg, et al. 2007). Consequently, central females maintain higher reproductive rates, elevated reproductive steroid production, higher offspring survivorship, and higher dominance status relative to peripheral females (Emery Thompson and Wrangham 2006; Emery Thompson, Kahlenberg, et al. 2007; Kahlenberg et al. 2008). We, therefore, predicted that high energy intake by central females should lead to relatively high C-peptide levels during lactation.
Urine sampling and analysis
Urine samples were collected noninvasively from chimpanzees as a routine part of daily follows. Whenever possible, researchers collected first-morning voids by placing plastic underneath nested individuals. Further sampling was conducted opportunistically throughout the day, and samples that could not be collected directly on plastic were pipetted from understory vegetation. Contamination by soil and feces was avoided, and any particulates in the sample were pipetted out after settling and prior to freezing.
During the study period (1998–2009), the community consisted of 40–50 chimpanzees. We sampled 17 adult females during the period of lactational amenorrhea, comprising data for all or part of 32 interbirth intervals (1–4 per female, Table 1). Since chimpanzees range widely and females frequently travel alone or in small groups, it was not possible to sample all individuals consistently from one month to the next. The first 2 years of lactation were sampled most intensively, with a total of 285 female-months from 17 individual females. C-peptide levels were also determined during periods of sexual cycling for comparison with lactation.
|Female (Subject IDs)||Estimatedyear of birth||Lactation periods (months) sampled||Urine samples|
|Female (Subject IDs)||Estimatedyear of birth||Lactation periods (months) sampled||Urine samples|
CP, C-peptide; E, estrone conjugates; PdG, pregnanediol glucuronide
Urine samples were assayed for metabolites of estrogens and progesterone (E1C and PdG, respectively), as well as C-peptide of insulin, though some sample volumes were insufficient to assay all hormones. Estrogen and progesterone are key products of the ovaries, and their relative levels predict conception success in cycling chimpanzees (Emery Thompson 2005). Increasing levels of these hormones are indicative of the gradual resumption of postpartum fecundity (Howie and McNeilly 1982; Wood 1994). The C-peptide is a protein cleaved from the proinsulin molecule during conversion to insulin, thus produced on an equimolar basis with insulin (Rubenstein et al. 1969; Melani et al. 1970). A variety of studies have utilized C-peptide as a measure of energetic condition in primates, validating it against various measures of dietary quality, habitat-wide food availability, body composition, activity levels, and individual caloric consumption estimates (Sherry and Ellison 2007; Emery Thompson and Knott 2008; Emery Thompson et al. 2009; Harris et al. 2010; Georgiev et al. 2011; Girard-Buttoz et al. 2011; Higham et al. 2011). Importantly, although insulin, and thus C-peptide, levels are expected to increase in proximate response to energy intake, a number of studies have demonstrated that urinary C-peptide varies in concert with change in body mass in a manner that is inconsistent with energy intake alone (Valeggia and Ellison 2004; Deschner et al. 2008; Girard-Buttoz et al. 2011). Others have documented that even during high energy intake, C-peptide levels decrease under conditions of high energy expenditure (Emery Thompson et al. 2009; Higham et al. 2011; Bergouignan et al. 2012). These points recommend C-peptide as a viable method for assessing the metabolic load of milk production (Ellison and Valeggia 2003).
E1C and PdG were determined via enzyme immunoassay, with reagents and protocols from the Clinical Endocrinology Laboratory at the University of California, Davis (CJ Munro), as described elsewhere (Emery Thompson 2005). C-peptide of insulin was determined via radioimmunoassay using kits distributed by Millipore (Billerica, MA), with samples diluted 1:2 in assay buffer. Average interassay coefficients of variation (CVs) of controls were 10.4% for C-peptide, 14.2% for E1C, and 14.8% for PdG. Intraassay CVs (average CV of duplicate determinations) were 7.0% for C-peptide, 5.0% for E1C, and 7.2% for PdG. All hormonal concentrations were standardized to creatinine (Cr), and overly dilute samples (Cr lower than 0.08mg/ml) were excluded.
During daily follows, teams of observers collected basic data on chimpanzee parties at 15-min intervals. These data indicate whether any females observed have a partially or fully tumescent sexual swelling and highlight the presence of new individuals (infants, immigrant females) and absence of expected ones (dependent offspring of females). We used these data to construct reproductive histories of females, including postpartum resumption of cycling, and infant birth dates. The onset of pregnancy was estimated by subtracting 230 days from the probable day of birth (Yerkes and Elder 1937; Martin et al. 1978; Shimizu et al. 2003). In some cases, estimates were refined by considering the last-observed sexual swelling period and/or hormonal indicators of pregnancy (positive hCG test strip or highly elevated E1C and PdG levels). Date of birth estimates for infants in this study are all considered accurate to within 1 month, based on observation records of females and size of infants at first identification. Females were considered to be in lactational amenorrhea from parturition until the first observation of a maximal sexual swelling. Though it is understood that during the later stages of this period not all mothers were necessarily producing milk, we refer to these females as “lactating” for brevity and because most of our analyses concern the early period of amenorrhea. If an infant died, the postpartum interval was considered in the analysis only up to the point of that death.
C-peptide and other hormonal measures were averaged for each female in each calendar month during lactational amenorrhea. To examine hormonal trajectories, where month or stage of lactation was a variable of interest, we constructed composite profiles of average C-peptide levels during each stage and truncated the analyses at the point when fewer than 6 birth intervals were sampled. For females who were sampled near the end of lactational amenorrhea, we constructed composite profiles comparable to those generated in Ellison and Valeggia’s (2003) study of humans, indicating each female’s C-peptide levels relative to the levels she experienced during the subsequent cycling period. In two cases, sampling was insufficient following resumption of cycling, so we then used that female’s average when cycling previously in the study. Though some females contributed multiple intervals to the dataset, there was extensive intraindividual variation, making it inappropriate to average separate intervals from the same female. To rule out pseudoreplication of individuals as a bias in our results, we also present C-peptide profiles in a reduced dataset generated from a single interval (the most complete) for each female. In our examination of C-peptide and reproductive hormones, we dealt with uneven sampling by applying a simple linear mixed models (LMM) procedure in which each female’s log-transformed monthly C-peptide concentration was a fixed covariate, either log-transformed E1C or PdG was the dependent variable, and individual identity was a random effect.
C-peptide dynamics over lactation
The first 1–2 years postpartum bear the strongest signature of lactation’s impact on female insulin production. C-peptide levels of lactating mothers remained relatively low and stable for approximately 6 months, with low variance across intervals (Figure 1). There was a net increase in C-peptide levels during the first 24 months following birth (R2 = 0.323, df = 23, P = 0.004), with the steepest increase occurring during the first year (R2 = 0.562, df = 11, P = 0.005, Figure 1). For those females who remained in lactational amenorrhea longer than 2 years, average levels of C-peptide were erratic and showed a net decrease back to levels similar to those of early lactation. This latter effect may be because the timing of cycle resumption varies substantially (range 6–97 months, mean = 41, N = 20), thus females who resume cycling early begin to drop out of the analysis in later stages. To gain a better perspective on this, we examine profiles retrospective to the onset of cycling in the following section. It should be noted that individual females did not experience a monotonic increase in C-peptide but showed substantial fluctuations in C-peptide levels over the course of nursing.
Effects on reproductive function
After controlling for variable representation of individuals, we found that the C-peptide levels of lactating females positively predicted their urinary estrogen (LMM estimate = 0.274, F = 63.189, df = 1, 354.1, P < 0.001) and urinary progesterone levels in a given month (estimate = 0.588, F = 98.000, df = 1, 413.6, P < 0.001, Figure 2).
For 18 postpartum intervals that were completed with the resumption of cycling and that had endocrine data proximate to that event, we aligned C-peptide profiles retrospectively from the month in which cycling resumed. This revealed a pattern of increasing C-peptide levels as the time of cycle resumption approached, reaching a peak approximately 4–6 months before cycle resumption (months −30 to −4, R2 = 0.374, df = 26, P = 0.001), falling just before cycling resumed (Figure 3).
Influence of resource access
We examined the influence of resource access on insulin dynamics in lactation by examining variance in C-peptide profiles exhibited by central parous females, those inhabiting high-quality core foraging areas, compared with peripheral parous females, those inhabiting resource-poor core foraging areas. The scope of this comparison was limited by the poor reproductive rates of the peripheral cluster females and the difficulty of locating and sampling them. However, we found that central females maintained higher C-peptide levels during the first year of lactation (Mann–Whitney U test, z = 1.998, Nc = 17, Np = 5, P = 0.046, Figure 4a,4b). There were not enough data to test whether this difference persisted after the first year. This result held true in the smaller dataset generated by using 1 interval per female (z = 2.208, Nc = 8, Np = 4, P = 0.027).
In this first physiological study of lactation in wild apes, we tracked changing insulin dynamics via urinary C-peptide over the course of lactation. Human and animal models indicate that relatively low insulin production during early lactation is reflective of metabolic shifts to accommodate the energetic requirements of milk production (Flint et al. 1979; Jones et al. 1984; Vernon 1989; Tigas et al. 2002). In our study, C-peptide levels of lactating chimpanzees remained relatively low and stable for the first 6 months of lactation, followed by a significant longitudinal increase through the first 1–2 years. This metabolic shift suggests that, despite the very long interbirth interval of chimpanzees and prior observations that weaning is not complete until approximately age 4-5 (Clark 1977), the intense energetic burden of lactation is concentrated in the first 1–2 years. Indeed, approximately one-third of mothers in our study had resumed cycling by the end of the second year.
Growth studies of chimpanzees confirm that the first 2 years of infant life are the most energetically demanding. These years encompass the highest relative gain in body mass (Leigh and Shea 1996; Marzke et al. 1996), the peak velocity in body length growth (Hamada and Udono 2002) and the majority of postnatal brain growth (Leigh 2004). We caution that these data derive from captive chimpanzees, who tend to have accelerated growth trajectories relative to their wild counterparts. However, the few data available on infant development in the wild also indicate an attenuation of infant demand during the first 2 years. In her study of 6 Gombe chimpanzee infants, Clark (1977) observed that, while nipple contacts continued until age 5, a marked decline in suckling frequency occurred after the first 6 months (corresponding to the upsurge in C-peptide in the mothers in our study), followed by a decline in suckling bout duration after 18 months. In a study of 5 different Gombe infants, peak nipple contacts occurred at approximately 6 months, followed by increasing rates of rejection by the mother and dramatic declines in ventro-ventro contact over years 1 and 2 (Rijt-Plooij and Plooij 1987). First consumption of solid foods has been observed at 4–5 months (Rijt-Plooij and Plooij 1987), a time when the majority of primary dentition is intact (Conroy and Mahoney 1991, data for captive chimpanzees, though Smith et al. 2010 found that crown formations for wild chimpanzees did not diverge substantially). Other data are consistent with some infants being nutritionally independent by 2 years. Although atypical, birth intervals as short as approximately 2 years (with the previous infant surviving) have been reported occasionally from wild chimpanzee populations (Emery Thompson, Jones, et al. 2007). A study of West African chimpanzees reported that orphaned infants of approximately 2 years of age were able to survive for prolonged periods in the absence of a breast-feeding surrogate (Boesch et al. 2009).
However, the interval to cycle resumption was longer than 2 years for the majority of females, with about one-third taking more than twice that time. Mean C-peptide levels in the aggregate declined after 2 years. Examination of individual profiles indicates that in some cases this was because those females with long periods of amenorrhea maintained low C-peptide levels from early stages of lactation, whereas in others, levels of C-peptide declined as lactation progressed into later stages. Substantial fluctuations evident in most C-peptide profiles likely correspond with changes in the availability of energy in the environment for both mothers and infants. In particular, because temporal habitat variability affects not only the quantity but also the types of foods available, the ability of infants to obtain appropriate weaning foods is expected to fluctuate. These data suggest that, even as infant caloric demands wane, maternal depletion from the costs of lactation can have a long-lasting impact on the energetic and reproductive condition of many females.
Our results parallel findings from human females in the Argentine Toba population (Ellison and Valeggia 2003) in relating the trajectory of C-peptide production to the onset of cycling. Like humans, chimpanzee mothers showed a pattern of significantly increasing C-peptide levels at the end of lactation as cycle resumption approached, and increases in C-peptide levels predicted increases in ovarian hormone production. As in the human study, the C-peptide levels of chimpanzee mothers substantially exceeded the cycling baseline levels for several months prior to the onset of cycling. Ellison and Valeggia (2003) (Valeggia and Ellison 2009) proposed that substantially elevated insulin concentrations are required to initiate normal ovarian activity after a prolonged period of insulin resistance during pregnancy and lactation. More generally, this pattern suggests that females require a sustained period of positive energy balance for cycling to resume. This may be a slow process for many females. In the population at Gombe, lactating chimpanzees have been reported to focus on a moderately higher quality (more fruit-rich) diet than other females, though they appear unable to devote a larger portion of their time to foraging (Murray et al. 2009). The ability to recuperate energy balance may also be limited by other costs and constraints of infant and juvenile care and by the increased costs of scramble competition for mothers (Wrangham 2000; Pontzer and Wrangham 2006).
In a variety of animals, parental condition and/or the availability of resources in the environment affect not only the degree of investment in parental care but also the relative trade-offs with future reproductive success. For example, in experimental manipulations of clutch size or food availability in birds, relative energy availability may affect egg quality or clutch size and can also have other downstream effects on fitness, such as through parental survival and the latency to produce a future clutch (Martin 1987). Accordingly, recent studies of mammalian mothers indicate that differences in the relative metabolic load of lactation, that is the energetic costs of milk production relative to the total energy budget of the mother, impacts future reproductive ability. This is because mothers ultimately draw on their own energy stores to subsidize lactation costs that cannot be met by increased intake. Thus, periods of lactational amenorrhea, and consequent birth intervals, of well-fed Toba women are considerably shorter than those of African hunter–gatherer women, despite the fact that women in both populations breast-feed at a similar intensity (Valeggia and Ellison 2001, 2004). Similarly, captive baboons with high energy intake during early lactation have a shorter interval to next conception while also spending more energy on milk production (Rosetta et al. 2011). Among marmoset monkeys bearing twin offspring, low body mass predicts greater loss of body fat during lactation, increased maternal morbidity, and lower likelihood of being fertile in the subsequent breeding season (Tardif et al. 2001). Among annually breeding phocid species, smaller mothers wean offspring earlier, but they do so because they apparently lack the energy reserves to sustain lactation any longer without compromising their own survival. Smaller mothers consequently produce smaller weanlings and are less likely to return to breeding condition in the following year (Arnbom et al. 1994; Fedak et al. 1996; Arnbom et al. 1997; Mellish et al. 1999). We therefore predicted that chimpanzee mothers with greater resource access would experience a lower metabolic load of lactation. Consistent with this prediction, mothers occupying higher quality foraging areas maintained higher C-peptide levels during early lactation than mothers in lower quality foraging areas. This difference in metabolic response to lactation may contribute to the marked differences in birth rates and offspring survival among these 2 groups of females (Emery Thompson, Kahlenberg, et al. 2007).
A drawback to our study is that, consistent with the community’s composition, most intervals in our dataset came from young but experienced mothers (i.e., parous females under the age of 30). Our dataset also underrepresented long birth intervals, particularly during the early lactation period, because long intervals are necessarily more rare, and those which were completed had often begun prior to our endocrine sampling. Thus, until our sample is extended, we are limited in our ability to examine some potentially interesting sources of variation in C-peptide profiles during early lactation, and we cannot reliably test whether the duration of amenorrhea is influenced by insulin dynamics during the earliest stages of lactation.
Taken together, our results provide important new insights on the interbirth intervals of chimpanzees, which are among the longest of any mammal. Our data, and those on the development of infant independence, suggest that the infant demands on maternal energy budgets begin to decline at approximately 6 months postpartum and that the period of significant energy transfer via lactation occurs for approximately 2 years. This accounts for only a fraction of the typical duration of lactation amenorrhea (~3–4 years, Wallis 1997) and of the total birth interval (~5–7 years, Emery Thompson, Jones, et al. 2007). However, our study also demonstrates that females need to reach and sustain a positive energy balance for several months in order for reproductive cycling to recommence. This situation is mirrored during the cycling period, where it has been found that poor or highly fluctuating fruit availability leads to long fertility delays (Emery Thompson and Wrangham 2008). Thus, we conclude that birth rates in chimpanzees are not limited by the physiological effects of infant suckling or even the daily energy costs of milk production, per se, but by the ability of mothers to extract enough energy from their environment to regain physical condition after a prolonged period of high expenditure.
National Science Foundation (Grant # 0849380); Leakey Foundation; American Association of Physical Anthropologists; Uganda Wildlife Authority; Ugandan National Council for Science and Technology; Makerere University Biological Field Station; Institutional Animal Care and Use Committees at Harvard University; and University of New Mexico.
For maintaining daily data collections over the long-term project, we thank our field assistants, the late J. Barwogeza, C. Katongole, F. Mugurusi, the late D. Muhangyi, the late C. Muruuli, and P. Tuhaiwe, James Kyomuhendo, Solomon Musana, Tweheyo Wilberforce, Sunday John, Chris Irumba, and field managers: Michael Wilson, Carole Hooven, Katherin Pieta, Kim Duffy, Alain Houle, and Emily Otali. Erin Fitzgerald provided valuable assistance in the laboratory.