RESEARCH | REPORTS PARASITOLOGY 70% of the population, the most common being hookworm (56%) and A. lumbricoides (15 to 20%) (11, 17). In both animal and human studies, parasites Helminth infection, fecundity, and have been shown to influence host reproduction via sexual behavior, brood or litter size, offspring age of first pregnancy in women size, incubation periods, conception rates, and pregnancy loss (18–22). In most cases, parasitism Aaron D. Blackwell,1,2,3* Marilyne A. Tamayo,4 Bret Beheim,2,5 reduces host reproduction by compromising Benjamin C. Trumble,1,2,3,6,7 Jonathan Stieglitz,2,5,8 Paul L. Hooper,2,9 reproductive organs or reducing energy budgets Melanie Martin,1,2,3 Hillard Kaplan,2,5 Michael Gurven1,2,3 (14, 23). However, among Tsimane adults, mor- bidity from intestinal helminth infections is low, Infection with intestinal helminths results in immunological changes that influence particularly for A. lumbricoides. Controlling for co-infections, and might influence fecundity by inducing immunological states affecting age and co-infection in our sample, hookworm conception and pregnancy. We investigated associations between intestinal helminths and infection is associated with slightly lower body fertility in women, using 9 years of longitudinal data from 986 Bolivian forager-horticulturalists, mass index (BMI) (generalized linear model, 2 experiencing natural fertility and 70% helminth prevalence. We found that different species b = –0.77 kg/m , P < 0.001) and hemoglobin (b = of helminth are associated with contrasting effects on fecundity. Infection with roundworm –0.19 g/dl, P =0.005),whereasA. lumbricoides is 2 (Ascaris lumbricoides) is associated with earlier first births and shortened interbirth intervals, not (b = –0.34 kg/m , P = 0.180; b = –0.07 g/dl, whereas infection with hookworm is associated with delayed first pregnancy and extended P = 0.413). However, helminth infection is also interbirth intervals. Thus, helminths may have important effects on human fertility that reflect associated with reductions in other infections, such physiological and immunological consequences of infection. as the intestinal protist Giardia lamblia (11). We hypothesized that intestinal helminths might in- crease fecundity because the associated immuno- ysregulated immune function, and in par- immune etiology (13); and also resemble the T cell logical changes, resembling those occurring during ticular autoimmune disease, has negative patterns that occur during pregnancy. pregnancy, modulate inflammatory responses that impacts on virtually every aspect of fecun- In humans, some helminth parasites can di- might otherwise impair fertility. dity, including ovarian function, implan- rectly infect the reproductive organs or the fetus; By using Cox proportional hazards models, we D Wuchereria tation, and pregnancy loss (1, 2). Healthy for example, the filarial roundworm, tested whether helminth infection was associ- pregnancy is also associated with shifts in im- bancrofti, can cause elephantiasis of the genitals ated with changes in birth spacing for 561 multi- mune responses. During the luteal phase of the (14). Animal studies have also examined life his- parous women and the age of first pregnancy menstrual cycle, regulatory and type 2 (TH2) T tory changes associated with parasitism (15). Yet (AFP) for 425 nulliparous women (24). Con- cell responses increase (3). If conception occurs, there are few data on the effects of intestinal sistent with our hypothesis, compared to being these shifts continue through pregnancy (4)and helminth infections on human fecundity, fer- uninfected, A. lumbricoides infection was asso- help to suppress type 1 (TH1) T cell responses, tility, or birth spacing. We prospectively exam- ciated with an earlier AFP hazard ratio [(HR) = increasing maternal tolerance of an immunolog- ined the effect of helminth infection on the 3.06, confidence interval (CI) 1.91 to 4.91, P <0.001 ically distinct fetus (3). Because pregnancy is fecundity of women. We used 9 years of longi- (Fig. 1 and Table 1)] and an increased hazard of both affected by and alters immunity, parasites tudinal data collected on 986 Tsimane forager- pregnancy under age 32 years (at age 20: HR = 1.64, that result in systemic immunological changes horticulturalist women living in the Amazonian CI 1.16 to 2.33, P = 0.005). This association might be expected to affect fecundity by alter- lowlands of Bolivia (table S1). Tsimane are pre- declines with age (interaction between A. lum- ing the host’s immune responses. Helminths, dominantly a natural fertility population, with bricoides and age: HR = 0.68 per decade, CI 0.51 such as hookworm (Ancylostoma duodenale or infrequent (<5% prevalence) use of pharmaceu- to 0.89, P = 0.006) and becomes significantly Necator americanus) and giant roundworm tical contraceptives and a total fertility rate of negative by the age of 46 years (HR = 0.62, CI (Ascaris lumbricoides), each infect 500 million nine births per woman (16). Helminths infect 0.38 to 1.00, P = 0.05). However, these late-life to 800 million people (5) and are associated with immunological changes, such that host helper Fig. 1. Associations between T cell populations generally shift from TH1 to- infection and likelihood of ward TH2 responses (6, 7) and the suppressive becoming pregnant. (A to C) activity of regulatory T cells increases to mod- Kaplan-Meier curves from Cox ulate both TH1 and TH2 responses (8, 9). These proportional hazard models shifts can alter susceptibility to other infectious (table S2), representing the time diseases, such as malaria (10), giardiasis (11), and to first pregnancy (A), and time tuberculosis (12); are associated with reductions to subsequent pregnancies in many diseases that have inflammatory or auto- at age 25 years (B) and age 40 years (C). Hazard ratios for conception associated with 1Department of Anthropology, University of California Santa Barbara, CA 93106, USA. 2Tsimane Health and Life History infection across ages are shown Project, San Borja, Bolivia. 3Broom Center for Demography, in (D). Colors indicate uninfected University of California Santa Barbara, CA 93106, USA. (dashed brown), infected with 4Department of Anthropology, University of Missouri, 5 hookworm (solid dark green), or Columbia, MO 65211, USA. Department of Anthropology, A. lumbricoides University of New Mexico, Albuquerque, NM 87131, USA. infected with 6Center for Evolutionary Medicine, Arizona State University, (solid yellow). Tempe, AZ 85287, USA. 7School of Human Evolution and Social Change, Arizona State University, Tempe, AZ, USA. 8Institute for Advanced Study in Toulouse, Toulouse, France. 9Department of Anthropology, Emory University, Atlanta, GA 30322, USA. *Corresponding author. E-mail: [email protected] 970 20 NOVEMBER 2015 • VOL 350 ISSUE 6263 sciencemag.org SCIENCE RESEARCH | REPORTS negative associations are outweighed by pos- ages (HR = 0.71, CI 0.58 to 0.86, P <0.001).A ical condition, education (a proxy of accultura- itive associations during early life, such that a woman chronically infected with hookworm tion), village location, season, and secular changes, woman with A. lumbricoides, projected across would be predicted to have three fewer children even though these variables do affect fertility her life span, would expect to have two more than an uninfected woman (Fig. 2). We found no [tables S2 and S3, also see (25)]. The results are children than a woman who was never infected interaction between infections, such that co- also not mediated by other comorbid infections (Fig. 2). infection is associated with the additive effects or illnesses (table S4). Twenty percent of infected In contrast, infection with hookworm was as- of hookworm and A. lumbricoides. women were given antihelminthic drugs during sociated with a delayed age of first pregnancy These results are not altered by controlling medical visits. Receipt of antihelminthics was (HR = 0.33, CI 0.20 to 0.54, P < 0.001) and a for other likely confounds affecting fecundity itself associated with a lower hazard of conceiv- reduced hazard of subsequent pregnancies at all or fecundity-altering behaviors, including phys- ing (HR = 0.75, CI 0.58 to 0.97, P =0.03);how- ever, neither controlling for treatment in models Fig. 2. Reproductive nor excluding these women appreciably altered careers predicted hazard ratios from infection with either hook- from Cox proportional worm or A. lumbricoides. The results are also hazard models, show- not driven by changing infection hazard during ing the expected dis- pregnancy. Although pregnancy is associated tributions of with an increased likelihood of hookworm infec- reproductive values tion, particularly in late pregnancy (table S6 and for hypothetical fig. S8), this relationship does not mediate the womenwithpersis- association between infection and conception tent parasite status hazards (24). Instead, it appears that hookworm- throughout life. Out- infected women occasionally clear their infections, comes include age at during which time they become pregnant, fol- first birth (A), interbirth lowed quickly by subsequent reinfection with intervals (B), age at last hookworm. Lastly, these associations are unlike- birth (C), age-specific ly to be caused by consistent differences between fertility (births/woman individual women (e.g., genetic pleiotropies) that per year) (D), median affect both fertility and risk of infection, because cumulative fertility over past parity is unrelated to likelihood of current time