AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 78519-526 (1989) Ecology and Ovarian Function Among Lese Women of the lturi Forest, Zaire PETER T. ELLISON, NADINE R. PEACOCK, AND CATHERINE T.AGER Department of Anthropology, Harvard University, Cambridge, Massachusetts 02138 (P.T.E., C.L.); Department of Anthropology, University of California at Los Angetes, Los Angetes, California 90024 (N.R.P.) KEY WORDS Reproduction, Luteal function, Progesterone, Nutrition, Salivary steroids (methods) ABSTRACT Ovarian function is examined in 35 Lese women inhabiting the Ituri Forest of northeastern Zaire over a period of 4 months through measurements of progesterone in saliva samples collected twice weekly. Ovulatory frequency is found to be only 56%on average, with a pattern of age variation similar to that observed in western women, though lower in level at each age. Average luteal progesterone levels of the Lese women are lower than those of Boston controls even if only ovulatory cycles are considered. Women with the poorest nutritional status, inferred from longitudinal weight changes and weight for height, show the greatest compromise of ovarian function, and the average ovulatory frequency of the whole sample declines in parallel with a period of weight loss over four months. It is suggested that low ovulatory frequency and luteal insufficiency contribute to the low fecundity of the Lese population and that nutritional status is likely to be one of the ecological factors modulating this effect. Our study site in the northern part of Zaire’s Ituri Forest lies in what has become known as the “African infertility belt,” a region stretching across the continent, along and north of the equator, characterized by isolated pockets of extremely low fertility and high rates of secondary sterility (Belsey, 1976). Ordinarily these pockets have been attributed to a high incidence of tubal sterility in women as a consequence of gonorrhea infection, although this explanation h a s been hard to verify. The Lese of our study area show several features typical of the African infertility belt. Among 119 postmenopausal women the average number of live births is only 2.4, with 37%reporting no live births and 10%reporting only one (Bailey, unpublished data). Other features of individual reproductive histories, however, seem at variance with the tubal sterility hypothesis, such as women whose recent pregnancies prove them to be still fecund, yet with interbirth intervals much longer than would be expected given their observed and traditional nursing behavior. 0 1989 ALAN R. LISS, INC In this report we present evidence that sheds new light on Lese fertility. I n particular, we show that suppressed ovarian function among Lese women may lead to a lower monthly probability of conception than is observed among western women. Further, we show that one of the factors associated with such ovarian suppression in this population is nutritional status. Together these observations constitute a n important addition to our knowledge of the ecology of human reproduction. We have published elsewhere the results of an initial study of ovarian function among the Lese and Efe, conducted in the summer of 1983 (Ellison et al., 1986). This initial study happened to occur at the end of a pronounced hunger season during which the average Lese woman lost 4.25 kg in 6 months. Based on measurements of salivary progesterone from 25 women with a history of recent infertility, we estimated that only 48%of our subjects ovulated during the month of observaReceived May 22, 1987; revision accepted April 29, 1988. 520 P.T.ELLISON ET AL. tion. Because our data were limited, however, we were unable to determine whether this low ovulatory frequency was a consequence of the period of weight loss or represented a characteristic feature of the population. Encouraged by the methodological success of the initial study, a much more extensive sample collection was undertaken from July to November 1984, during the period of the year immediately following the peanut harvest and honey season, when the nutritional status of the population is ordinarily at its highest. MATERIALS AND METHODS Once again our principal objective was to measure progesterone in samples of saliva. Progesterone is secreted in quantity in the second, luteal half of the cycle, ordinarily as a consequence of ovulation. Although luteinization of unruptured follicles is possible (Stanger and Yovich, 1984),a significant elevation of luteal progesterone is generally considered as reliable evidence of ovulation (Riad-Fahmy, 1984; Walker et al., 1979). Adequacy of luteal progesterone production by the corpus luteum appears to be a clinically important determinant of the probability of implantation and early viability of the trophoblast (Hamilton et al., 1987; Ftiddick et al., 1983). Low levels of luteal progesterone and/ or shortened periods of progesterone secretion, collectively referred to as luteal phase defects or luteal insufficiency, are chronically associated with infertility (DiZerega and Hodgen, 1981; Annos et al., 1980). Saliva samples are easily collected under field conditions, and the progesterone concentrations measured in saliva allow comparison of ovulatory frequency and luteal sufficiency with data from other populations (Ellison, 1988). From July through November 1984 saliva samples were collected twice weekly from all Lese women in the study area past menarche. Of that sample, all women who were menstruating during the study period, hereafter referred to as subjects, were included in the analyses presented here. Three subjects became pregnant during the sampling, others were occasionally absent from the study area. None of the subjects was lactating during the sampling period or had children under the age of 2 years. The total sample for these analyses thus includes 89 cycles collected from 35 women, or a n average of 2.6 cycles per woman. Selected aspects of the reproductive histories of these women are included in Table 1. Subjects were visited twice a week, in the early morning, when a sample of 4-5 ml of saliva was collected in a polystyrene tube. Each woman was questioned at each visit about recent menstruation. A colored bead was added to a copper wire bracelet on each occasion a s a simple quid-pro-quo and as a n individual tally of samples given. Back in camp 50 pl of sodium azide solution was added to each sample as a preservative, bringing the concentration to approximately 0.1%.Once a month all subjects were weighted to the nearest 0.5 kg using a portable scale. The samples were stored at ambient temperatures for up to a month before being frozen at -20°C. Progesterone levels in saliva are stable under these conditions when preserved with sodium azide (Ellison, 1988; Ellison et al., 1986). In the laboratory, 2 ml of saliva from each sample were extracted in ether and added to a radioimmunoassay of progesterone. Details of this assay procedure are published elsewhere (Ellison and Lager, 1986;Ellison et al., 1986). Modifications of the assay since our initial project in Zaire have improved the sensitivity and precision of our measurements, the smallest measurable amount now being 25 pmol/L and the coefficient of variation of sample determinations being less than 10%. The time series of data from each individual subject was analysed for evidence of progesterone peaks using the program PULSAR developed by Merriam and Wachter (1982) at NICHHD (Fig. 1). PULSAR extracts a smooth baseline from the data series by a n iterative application of a robust running averages procedure and identifies peaks as significant departures from this baseline. Identified peaks are excluded from the next iteration until no further peaks are detected. With this method, all cycles except four could be categorized as ovulatory or not; TABLE 1. Selected aspects of the reproductive histories of Lese subiects Average interval since last birth for Average Age group % % parous women No. of (years) N childless unmarried (months) live births 4 2 2 11 23-35 19 236 5 54.5 26.3 20.0 27.3 5.3 0.0 39.8 56.6 132.7 0.7 1.6 2.8 521 OVARIAN FUNCTION AMONG LESE WOMEN 600 + PROGESTERONE 0 BASELINE 400 1 500 P 400 > 300 5 200 ; m 0 20 40 60 80 100 120 DAYS Fig. 1. Example of the use of PULSAR to analyse the time series of progesterone data from a single subject. Solid squares represent observed progesterone values, open squares represent the extracted baseline, and asterisks represent the significant progesterone peaks identified by the program. the four ambiguous cycles were examined individually and subsequently included in the ovulatory group. The method of assaying salivary progesterone was originally developed by Walker and colleagues at the Tenovus Cancer Institute in Wales (Walker et al., 1979). Results from assays performed in our laboratory at Harvard correspond closely with theirs (cf. Ellison et al., 1986). Samples collected from cycling female researchers in the field also demonstrate profiles comparable to the Boston and Welsh control values. In Figure 2, profiles from three study subjects are depicted illustrating the correspondence between salivary progesterone measurements and various reproductive states. Statistical analyses of group differences in progesterone levels are performed on logtransformed data unless otherwise noted, to normalize distributions. Comparisons of ovulatory frequency are by chi-square statistic unless otherwise noted. RESULTS Although the 1984 field season had been planned to fall during a period of relative food abundance, food supplies were not, in fact, as bountiful as usual. Early and extended rains the previous December, January, and February had once again interfered with the process of clearing and burning new gardens, and many Lese families were compelled to rely solely on a replanting of the previous years plots. In addition, the honey season in June and July was the worst for 100 o 0 1'0 . 20 . 30 40 . 50 . 60 , 70 , 80 DAYS Fig. 2. Sample salivary progesterone profiles of Lese women in different reproductive states. Solid squares trace the profile of a postmenopausal Lese woman. The rising profile traced by the solid circles is that of a woman who became pregnant during the sampling and who subsequently delivered a healthy baby girl. The open circles trace the profile of a woman who bore twin boys prematurely at 6 months gestational age, both of whom died within 12 hours. Notice that in the absence of lactation normal cycles resume almost immediately. Fig. 3. Distribution of net weight changes between July and November 1984 in Lese subjects. (The three subjects who became pregnant during the study are not included.) many years, so that this normal infusion of calories from the forest was virtually absent for the Lese. Although by no means as severe as the situation in the spring of 1983, there was a steady decline in the average weight of the study subjects between July and November, the average net change being a loss of 0.95 kg (Fig. 3). 522 P.T.ELLISON ET AL. Once again ovulatory frequency was found to be quite low: 54% weighting all cycles equally, 56% weighting individual women equally. Subdividing the subjects into three age categories-522 years, 23-35 years, and 2 3 6 years-reveals a distinct age pattern (Fig. 4): young women under 23 years of age ovulate a mere 33% of the time on average versus an ovulatory frequency twice as high among older women aged 23 years and older (P< .05). The slight decline in ovulatory frequency between subjects aged 23-25 years and subjects aged 36 years and over is not statistically significant, as is often reported for western women (Doring, 1969). Although comparable in shape to the age curves of ovulatory frequency available from western populations, the ovulatory frequencies of the Lese women are substantially lower than those of similarly aged women in the west. For example, the ovulatory frequency we found in a sample of 25 college undergraduates aged 18-22 years using the same method was 58%(P< .01 compared to Lese sample), and that among adult women was 100%(P< .005 compared to Lese sample; Ellison et al., 1987, Fig. 4). Not only is the frequency of ovulation low among the Lese women, the composite progesterone profile as a whole is low compared to that of Boston women previously found comparable to published standards from the Tenovus Institute (Riad-Fahmy, 1984; Ellison et al., 1986). To some extent the low composite profile results from including progesterone levels from anovulatory Lese cycles. Even if ovulatory cycles alone are considered, 16-22 23-35 36-50 AGE GROUP Fig. 4. Ovulatory frequency by age for the Lese subjects compared to that of similarly aged Boston women. however, the Lese composite profile is low compared to that of Boston women (Fig. 5). Nor does the inclusion of late adolescent cycles in the Lese sample account for the difference from the Boston sample. Significant differences can in fact be found between the average luteal progesterone level of Boston women and that of any of the following: the entire Lese sample, only Lese women 2 2 3 years of age, only ovulatory cycles of Lese women, and only ovulatory cycles of Lese women 1 2 3 years of age (P< .05, one-tailed t-tests, Fig. 6). 500 400 -mccNlw.s + OWLATORY + AULESE 300 200 100 0 -30 -20 -10 0 DAYS BEFORE MENSES Fig. 5. Average salivary progesterone profiles of the entire Lese sample, the subset of ovulatory Lese cycles, and a sample of Boston women aged 23-35 years (Ellison and Lager, 1986). Fig. 6. Comparison of average luteal level of salivary progesterone (+SE) among five sample groups: Boston women aged 23-35 years, all Lese subjects, only ovulatory cycles of Lese subjects, only cycles of Lese subjects over age 23 years, and only ovulatory cycles of Lese subjects over age 23 years. OVARIAN FUNCTION AMONG LESE WOMEN The women sampled in the initial study in 1983 correspond in age to the middle age group of women in the present study. The contrast between the 48% ovulatory frequency in 1983 and the 69% ovulatory frequency in this age group in 1984 may thus reflect the more serious nutritional situation in 1983, or it may represent differences in the frequency with which reproductively impaired women were included in the two samples. We will return to the question of sample bias in the discussion. The more extensive data collection in 1984 allows us to investigate the relationship between nutritional status a n d ovarian function, however, within the Lese sample. I n particular, z 2.0 Kg < 2.0 Kg ’ 200 NET WEIGHT CHANGE (kg) Fig. 7. Ovulatory frequency and average peak progesterone levels (SE) in L e e subjects (excluding three subjects who became pregnant during the study) subdivided by net weight change July to November 1984. 523 three analyses provide evidence of a positive relationship between these two variables. First, the subjects can be subdivided on the basis of their net weight change, July to November (Fig. 7; the three subjects who became pregnant during the study are excluded). Ovulatory frequency is slightly, but not significantly, lower among those women who lost more than 2 kg than among those who lost less or gained weight. The average peak progesterone level (ovulatory cycles only) attained by the women who lost more than 2 kg is significantly lower than that attained by women who lost less or gained weight (P< .05, one-tailed t-test). Second, the subjects can be subdivided by weightfor-height at the start of the study as a n index of current nutritional status (Fig. 8). Once again ovulatory frequency (nonsignificantly) and peak progesterone levels (P< .05, one-tailed t-test) are lower in the subjects with poorer nutritional status. These results indicating that women with poorer nutritional status also suffer impaired ovarian function do not appear to be artifacts of age differences between the groups a n a lysed (see “Discussion”). Finally, monthly ovulatory frequency declined steadily over the course of the study, from 63%in JulyAugust to 30%in October-November (Fig. 9, P < .001; significance of regression of ovulatory frequency against time, P < -005). DISCUSSION The results of the current study of Lese ovarian function once again indicate that ovulatory frequency and progesterone levels 500 0 -2 % OVUIATORY + PEAKPRCG 0 400 h ; z 8 t- v1 CI 300 p Y 2 200 5 0.3 z 0.3 WEIGHTIHEIGHT (kglcrn) Fig. 8. Ovulatory frequency and average peak progesterone (fSE) in Lese subjects subdivided by weight-forheight at the start of the study. Fig. 9. Monthly ovulatory frequency of Lese subjects over the course of the study. 524 P.T.ELLISON ET AL. are low compared to western women. When adjusted for age the ovarian suppression does not appear to be as great as that observed in 1983 when the population was under more severe nutritional stress. Close examination of the 1984 data suggests that those women in the population with the poorest nutritional status initially and those who lost more weight during the study show the greatest compromise of ovarian function, particularly as regards luteal progesterone levels. Neither of these results appears to be a n artifact of the distribution of young women, whose ovarian function is much lower than that of older women, between the groups analysed. There is no significant difference in average age between the subjects with a weight-forheight index less than 0.3 and those with a n index over 0.3 (29.2 k 2.7 vs. 26.2 & 1.7 years, mean k SE). The 11subjects in the under 23year age group, who show a significantly lower level of ovarian function, are evenly distributed between the two groups, six in the group with a n index over 0.3 and five in the group with a n index under 0.3. When women are grouped by weight change, the group of women who lost less than 2 kg is significantly older on average than the group of women who lost more than 2 kg (29.3 k 2.0 vs 23.8 k 1.7 years, P < .05, two-tailed t-test). The youngest age group is, however, also evenly split in this case, the age difference being due to the distribution of older women among whom no difference in ovarian function was detected. These results together imply that Lese ovarian function covaries with nutritional status, in particular with the mild yet progressive weight loss associated with a period of relative food shortage. Associations of this kind have also been reported for western women. I n a study of nine normalweight women aged 20-29 years undergoing voluntary weight loss through dieting, Pirke et al. (1985) found no evidence of diminished gonadotropin secretion after 6 weeks, but significantly lowered levels of gonadal steroids, especially progesterone. Bullen et al. (1985) found that among women undergoing a prescribed exercise regimen, those who were allowed to lose weight showed more evidence of impaired ovarian function than those who were not allowed to lose weight. In our own studies we have found that Boston women undergoing voluntary weight loss through dietary restriction show lower ovulatory frequency and lower luteal pro- gesterone levels than do similarly aged controls of stable weight (Lager and Ellison, 1987). The suppression of ovarian function among such women was even more pronounced in the cycle following that in which weight loss occurred than in the weight loss cycle itself (Fig. 10). The results of our current study among the Lese are not out of line with the findings of these previous studies. Our results from the Lese, however, constitute the first evidence of a covariance of nutritional status and ovarian function as part of the natural ecology of a noncontracepting population. Although the covariance of nutritional status and ovarian function among the Lese is comparable to that observed in the west under more artificial conditions, it is not yet clear whether nutritional status alone can account for the dramatic difference in baseline level of ovarian function between the Lese sample and western women of comparable age. Two other important possibilities are worth noting: first, that the sample of Lese women studied is biased toward the inclusion of women of impaired ovarian function; and second, that the suppressed ovarian function of Lese women is a chronic feature of their reproductive ecology,perhaps reflecting the impact of other ecological variables. Unlike the initial 1983 study, in which subjects were chosen on the basis of a recent history of infertility, all menstruating women within the study area were included in the 500 1 -=-ccNrFa + WEIGKTLOSS + POST LOSS fi 0 4 ~ 3 0 I -20 -10 0 DAYS BEFORE MENSES Fig. 10. Comparison of the average salivary progesterone profiles from 19 cycles of normally menstruating women, 14 cycles of weight loss in dieting women, and 9 cycles which followed a month of weight loss in dieting women (Lager and Ellison, 1987). OVARIAN FUNCTION AMONG LESE WOMEN present study in a n attempt to minimize selection bias. Nevertheless, some overrepresentation of women of diminished fecundity in our sample must be assumed. I n any natural fertility population that is heterogeneous with respect to female fecundity, women of impaired fecundity will spend a greater proportion of their lives in the pool of menstruating women than fully fecund women. This tendency will be exaggerated to the extent that some women wean their infants as a consequence of becoming pregnant instead of the other way around, as is reported by some Lese women. We fully acknowledge that such a n overrepresentation of low fecundity women in our Lese sample is likely. The noteworthy result of our research is that such diminished fecundity is associated with suppressed ovarian function, and in particular with low luteal progesterone, which may therefore be a n important contributing factor to the low fertility of the population. It is possible that venereal disease may produce such impaired ovarian function. In particular, it is possible for gonorrhea infections to be so severe that the ovary becomes scarred and ovulation is prevented (Svensson et al., 1983), but this result is rare in western clinical experience and is always accompanied by acute symptoms of pelvic inflammatory disease. Furthermore, in our sample evidence of luteal suppression persists even if anovulatory cycles are excluded. These considerations suggest that the suppressed ovarian function observed is unlikely to be a simple result of venereal disease. It is also possible t h a t the suppressed basal level of ovarian function in the Lese sample is a chronic feature of their reproductive physiology, deriving from genetic, developmental, or environmental causes other than poor nutritional status. No evidence exists for strictly genetic differences in reproductive physiology between populations, and the possibility of such a difference distinguishing the Lese seems unlikely. A proper investigation of this question, however, would require quite different data from what are available to us. Development a l differences in ovarian function have been demonstrated among western women, with late-maturing women having a low trajectory of ovulatory frequency by age relative to early maturers (Apter and Vikho, 1983). The same factors that cause the Lese 525 to be a relatively late-maturing population, with a n average menarcheal age of about 16 years, may result in suppressed ovarian function throughout the average woman’s life. Finally, other ecological factors, in particular heavy workloads and high disease burdens, may combine with marginal nutritional status to suppress Lese ovarian function. Even a mild exercise regimen, if pursued regularly, is capable of suppressing progesterone profiles and lowering ovulatory frequency to a degree comparable to that observed in the Lese subjects. Boston women who run a n average of 12.5 miles a week for exercise show salivary progesterone profiles much like those of the Lese with comparable ovulatory frequency and average luteal levels (Ellison and Lager, 1986). Fairly heavy physical labor, such as cultivating the gardens, or carrying firewood and children, is a regular feature of the lives of Lese women. This possibility has been the focus of a third field season in 1987 conducted by one of us (N.R.P.), which we hope to report on soon. In summary, we have shown that menstruating Lese women are characterized by suppressed ovarian function, particularly lower luteal progesterone levels, relative to western women of the same age; we have argued t h a t this suppression of ovarian function is a likely contributor to the low fertility of the Lese population; and we have shown that within the sample of menstruating Lese women, level of ovarian function covaries with nutritional status. As well as adding to our understanding of Lese reproduction, these results illustrate the important contribution that the study of ovarian function can make to the broader subject of human reproductive ecology. ACKNOWLEDGMENTS We wish to thank Robert C. Bailey for the use of his demographic data, and the many members of our project who contributed to the collection of anthropometric data. Our special thanks to the Lese women of Malembi. This work was supported by NSF grant BNS83-19629 to Irven DeVore and Peter Ellison. LITERATURE CITED Annos T, Thompson IE, and Taymor ML (1980) Luteal phase deficiency and infertility: Difficultiesin diagnosis and treatment. Obstet. Gynecol. 55:705-710. Apter D, and Vikho R (1983)Early menarche, a risk factor for breast cancer, indicates early onset of ovulatory cycles. J. Clin. Endocrinol. Metab. 57r82-88. 526 P.T. ELLISON ET AL. Belsey MA (1976)The epidemiology of infertility: A review with particular reference to sub-Saharan Africa. Bull. WHO 54:319-341. Bullen BA, Skrinar GS, Beitins 12, von Mering G, Turnbull BA, and McArthur J W (1985) Induction of menstrual disorders by strenuous exercise in untrained women. N. Engl. J. Med. 3I2:1349-1353. DiZerega GS, and Hodgen GD (1981) Luteal phase dysfunction infertility: A sequel to aberrant folliculogenesis. Fertil. Steril. 35489-499. Doring GK (1969) The incidence of anovular cycles in women. J. Fkprod. Fert. [Suppl.] 677-81. Ellison PT (1988) Human salivary steroids: Methodological issues and applications in physical anthropology. Yearbook Phys. Anthropol., Yearbook Phys. Anthropol. 31:115-142. Ellison PT, and Lager C (1986) Moderate recreational running is associated with lowered salivary progesterone profiles in women. Am. J. Obstet. Gynecol. 154: 1000-1003. Ellison PT, Lager C, and Calfee J (1987) Low profiles of salivary progesterone among college undergraduate women. J . Adolesc. Health Care 8204-207. Ellison PT, Peacock NR, and Lager C (1986) Salivary progesterone and luteal function in two low-fertilitypopulations of northeast Zaire. Hum. Biol. 58:473-483. Hamilton CJCM, Evers JLH, and de Haan J (1987)Ovulatory disturbances in patients with luteal insufficiency. Clin. Endocrinol. (Oxf.) 26:129-136. Lager C, and Ellison PT (1987) Effects of moderate weight loss on ovulatory frequency and luteal function in adult women. Am. J . Phys. Anthropol. 72921-222. Merriam GR, and Wachter KW (1982) Algorithms for the study of episodic hormone secretion. Am. J. Physiol. 243:E310-E318. Pirke KM, Schweiger V, Lemmel W, Krieg JC, and Berger M (1985) The influence of dieting on the menstrual cycle of healthy young women. J . Clin. Endocrinol. Metab. 60:1174-1179. Riad-Fahmy D (1984) Salivary progesterone for investigating ovarian activity. Front. Oral Physiol. 5:llO-123. Riddick DH, Daly DC, and Rosenberg SM (1983) Progesterone and endometrial physiology in human infertility. In CW Bardin, E Milgrom, and P Mauvais-Jarvis (eds.): Progesterone and Progestins. New York: Raven, pp. 861-865. Stanger BM, and Yovich J L (1984) Failure of human oocyte release at ovulation. Fertil. Steril. 41:827-832. Svensson L, Mardh PA, and Westrom L (1983) Infertility after acute salpingitis with special reference to Chlamydia trachoma&. Fertil. Steril. 43:322-329. Walker RF, Read GF, and Riad-Fahmy D (1979) Radioimmunoassay of progesterone in saliva: Application to the assessment of ovarian function. Clin. Chem. 25: 2030-2033.