AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 127:105–113 (2005) Biological and Behavioral Determinants of Fertility in Tierra del Fuego J. Pascual,* C.E. Garcı́a-Moro, and M. Hernández Unitat d’Antropologia, Facultat de Biologia, Universitat de Barcelona, 08028 Barcelona, Spain KEY WORDS Chile human fertility; path analysis; proximate determinants; biodemography; ABSTRACT The reproductive history of 182 women in postreproductive life or near menopause from the Chilean part of Tierra del Fuego was traced back by means of familial interviews. These postmenopausal women represent the population since almost the beginning of the settlement, and their reproductive years were spent on the island. Path analysis was applied to analyze fertility determinants of these women and to propose a complex model of interconnections among factors. The reproductive history of these women is characterized by a long fertile span, a short childbearing period, and low fertility. Age at menarche is relatively late, and the age of the women at ﬁrst birth is mainly determined by their late age at marriage. The use of contraception is related to both spacing and stopping behaviors. The late age of women at marriage, the rhythm of conception, and practices of contraception are proposed as the main determinants of fertility in Tierra del Fuego. Am J Phys Anthropol 127: 105–113, 2005. © 2004 Wiley-Liss, Inc. An approximation to the analysis of fertility determinants in a human population is a multivariate study of the connections among several events of the reproductive histories of women. Correlation coefﬁcients are often used to estimate the connection between several factors, even though multivariate techniques such as hazards analysis are in use. Most studies of fertility in human populations choose a few factors estimating the correlation matrix; after that, a consistent explanation of results is written by the researchers. These results can be analyzed in an independent way, but they do not provide an overall view of the reproductive model of the population. One of the goals of this article is to design a complex model of interconnections that incorporate various factors considered in the reproductive history of women surveyed in Tierra del Fuego. This model should explain as much as possible the interconnections among factors, using the method of path analysis. Geneticists and breeders have long known and have been using the method of path coefﬁcients developed in 1920 –1921 by the pioneer population geneticist Sewall Wright. In last decades, the path analysis method has been used in population genetic studies and other anthropological sciences, especially in the estimation of both heritability and family transmission of characters (Salces et al., 2001). Moreover, the method of path analysis is an interesting procedure to build and evaluate models about connections among factors. Nevertheless, the use of path analysis in fertility studies of human populations is rare (Crognier, 1996; Westoff, 1990; Loebner and Driver, 1973) and, in general, in biodemography studies (Shen and Williamson, 1999; Gober, 1997; Clay and Vander Haar, 1993). Recently, Madrigal et al. (2003) concluded that heritability of fertility is moderated, and they used path analysis to show that the anthropometric biological characteristics of mothers explain small variability in a population’s fertility. Factors that better explain differences in the fertility of women in Tierra del Fuego should be found among other biological and sociocultural characteristics of the population. This biodemographic study of the population of the Chilean part of Tierra del Fuego attempts to establish a fertility model. A statistical method that allows us to integrate an important number of variables and to deﬁne both a consistent and a structured model was used. © 2004 WILEY-LISS, INC. POPULATION AND METHODS The Chilean province of Tierra del Fuego is situated at the southern tip of the South American continent (52.5–55° latitude), which conditions its environmental and climatic characteristics. The Chilean territory of Isla Grande de Tierra del Fuego was *Correspondence to: Joel Pascual, Unitat d’Antropologia, Facultat de Biologia, Universitat de Barcelona, Avda. Diagonal 645, 08028 Barcelona, Spain. E-mail: email@example.com Received 6 July 2003; accepted 12 February 2004. DOI 10.1002/ajpa.20065 Published online 31 August 2004 in Wiley InterScience (www. interscience.wiley.com). 106 J. PASCUAL ET AL. inhabited by the aboriginal Selk’nam populations (Garcı́a-Moro et al., 1997a) until the end of the 19th century, when a quick reduction in these populations and their near-extinction happened with the arrival of European expeditions. The colonizing population arrived, at the end of the 19th century, from other Chilean regions (particularly Chiloé) and diverse European countries, especially Britain and Croatia, but also Germany, Spain, and Italy. The majority of inhabitants live in Porvenir, the administrative capital of the province (2002 census: 6,904 inhabitants). From the beginning of the settlement, this community has been a semirural developing population, where ranching activity was predominant (Pascual et al., 2000). A survey was carried out in the austral spring of 2000 in Porvenir. The reproductive history of women was traced back by means of familial interviews, including personal information, data on stable unions, and biological characteristics of the women and information on their children. Women were selected randomly, and all neighborhoods and social classes of Porvenir are represented. The women surveyed are predominantly of Chilean origin, mainly from Tierra del Fuego, other locations of the Magellanic Region, and the Chiloé Islands. They show a semirural economy (most are from families dedicated to ranching activity), and they have a general low level of education (68% of the women studied less than 7 years of school, and only the 6.6% of the women had university studies). Interviewed women met two previous conditions: 1) to be more than 45 years old at time of survey (they were born between 1912–1956, so the population since almost the beginning of the settlement is represented), and 2) to be born in Porvenir or to have lived there since childhood or youth, so that their reproductive period was spent in Tierra del Fuego. Local authorities gave support to the researchers to obtain a census of women who ﬁt the study criteria. The possibility to subdivide women into several subgroups to analyze temporal trends of factors or changes in patterns exists, but the reduction of sample size increases the heterogeneity of variables; assuming this, path diagrams would be less consistent and, therefore, moot structures. Thus, we decided to keep together the whole group of surveyed women. The method of path coefﬁcients (Li, 1975) is a form of structured linear regression analysis with respect to standardized variables in a closed system. To apply this method, the ﬁrst step is to create a list of biological, demographic, and sociocultural factors that will be considered in the analysis. In this case, we made a special effort to choose the most important and direct factors a priori, and this selection was based on the knowledge of the population and factors suggested by the literature. After that, a structure concerning the interconnections among variables is assumed. To do this, factors were classiﬁed according to the time sequence in reproductive life of a woman, and ﬁnally, we decided on paths and directions of the connections. Low multicollinearity among variables is indispensable for including several factors in the same path. When the path diagram is done using correlations among factors, the mathematical analysis quantiﬁes with different coefﬁcients (path coefﬁcients) the importance of every connection established in the model. To work with correlations assumes that variables included in diagrams are normally distributed; however, most factors related to fertility are non-normal variables. In this case, the strength of results will depend on the factors’ involvement in women’s fertility. The path coefﬁcient is deﬁned as the ratio of the total variability that can be explained by only one factor when all the rest of the variables are unchanged (Li, 1975). In other words, a path coefﬁcient represents the net inﬂuence of a variable on another one. The residual (r) includes heterogeneity not explained by factors (not confounded with the correlation coefﬁcient). The squares of the path coefﬁcients measure the degree of determination by each factor (direct determination), but an indirect determination of each factor can also be quantiﬁed by means of the other variables considered (combined determination). Then the direct determination quantiﬁes the inﬂuence of variable A on variable B across the considered path. At the same time, A can inﬂuence B across other different paths, and this is quantiﬁed by means of the combined determination. The unexplained variability (1 ⫺ total determination) can be estimated when the total determination (deﬁned as the sum of both direct and combined determinations) is known. The correlation between two variables is the sum of all connecting paths between them. Then, a correlation can be decomposed into various components (path coefﬁcients), each corresponding to a path. The path coefﬁcient coincides with correlation coefﬁcient only when you consider one component. In summary, path analysis enables us to 1) hypothesize a complex causal structure, 2) draw a visual display of the model, 3) calculate the strengths of the various connections, and 4) insert their coefﬁcients into the causal diagram. Moreover, the method provides a general procedure to explore the indirect effects of a determining variable on a dependent variable in a multivariate path model, which seems more informative than partial correlation analysis (Loebner and Driver, 1973). On the other hand, whenever we assume a structure about the interconnections among a number of variables, it is unavoidable to involve some degree of subjective judgment: two investigators, given the same data on the same variables, may very well come up with two different understandings of the variables and propose two different path diagrams for analysis. Hence, the most important disadvantage of path analysis is the dependence on an a priori causal scheme (Loebner and Driver, 1973). 107 FERTILITY IN TIERRA DEL FUEGO TABLE 1. Abbreviations of variables in text Variable Abbreviation Variable Abbreviation Menarche Age at marriage Mean birth interval Age of mother at ﬁrst birth Age of mother at last birth Childbearing period M AAM MBI AMFB AMLB CP Years using contraceptives Years of fertile marriage Pregnancies Abortions Number of live births CONTRAC YFM P ABOR NLB However, path analysis requires that results must be consistent throughout the structure and compatible with the observed data on all variables involved in the structure (Li, 1975). Twelve variables of the reproductive life of women were initially considered in the analysis, and abbreviations used in text are expanded in Table 1. Among the diversity of schemes which explain interrelationships among variables, two path diagrams were selected, considering the proposed connections in the literature and the considerable percentages of variability that they explained. It is possible that some connections among variables suggested in the literature are not expressed in these particular path diagrams. This is because they are less quantitatively important than others in Tierra del Fuego, and the absence of arrows ensures no connection in the diagram. These unconsidered connections are included in indirect paths and residual components. We used SPSS 10.0 and Mathematica 4.0 to estimate statistical and algebraic values. The age at menarche and age at menopause were estimated by a retrospective method. The limitations of this method were described elsewhere (Madrigal, 1991). However, several studies conﬁrmed that women remembered their menarcheal age with a minor error (Bergsten-Brucefors, 1976; Tryggvadóttir et al., 1994). Therefore, a 0.5-year midpoint correction factor was added to the reported age, when estimating both mean age at menarche and menopause. RESULTS Main reproductive characteristics The main events that determine the reproductive life of postmenopausal women in Tierra del Fuego are shown in Table 2. The age at menarche and age at menopause were recently analyzed in Pascual et al. (2003). Their main conclusion is that women from Tierra del Fuego show a long fertile span, even though the mean age at menarche is relatively late. Moreover, 32.9% of women lost their fecundity due to pathological or voluntary surgery, and these women cannot be included in the sample of the fertile span. The natural fertile span, estimated as the difference between age at menopause and age at menarche, is 34.12 years; nevertheless, the childbearing period is more informative than the fertile span because this is the real period dedicated to reproduction. The childbearing period, estimated as the difference between the age of women at ﬁrst and TABLE 2. Main reproductive events in reproductive histories of women from Tierra del Fuego1 Variables Mean SD N Women’s age Age at menarche Age at marriage Age at ﬁrst birth Age at last birth Age at menopause Fertile span Childbearing period Birth interval Pregnancies Live births 62.06 13.85 23.03 23.66 32.11 48.38 34.12 8.89 3.69 3.77 3.31 11.35 1.59 5.22 5.39 6.29 4.85 4.86 5.53 2.75 2.15 1.91 168 162 159 156 155 99 73 147 142 168 168 1 All measures are estimated in years. last birth, is 8.89 years. Thus a long theoretical fertile span does not entail high fertility, because the practical reproductive span is short. In a multifactorial view, the number of live births in a woman’s reproductive life is determined by a few proximate variables (Bongaarts and Potter, 1983). Late age at marriage of women reduces by 10 years the theoretical period to bear children. The age of the mother at ﬁrst birth is much related to age at marriage. Both variables are very close, because intercourse before marriage is frequent in Tierra del Fuego, and in many cases marriage takes place after a period of cohabitation. On the other hand, the mean length of birth intervals is relatively long and the age of the mother at last birth is too early for a natural fertility population (Bongaarts and Potter, 1983). The extensive use of contraception and the importance of deﬁnitive sterilization are suggested by these results. Fertility is relatively low in the Chilean part of Tierra del Fuego. Postreproductive women experience an average of 3– 4 pregnancies, and they have an average of three live births. The total fertility rate (number of live births per woman) is 3.3 (including 7% of childless women), and the total marital fertility rate (number of live births per married woman) is 6.4 births per woman. In summary, this low fertility could be the consequence of a late age at marriage, a short childbearing period, and the use of contraceptives. Table 3 shows the correlation matrix between variables considered in the model. Path coefﬁcients were estimated from these values. Suggested connections among factors and their associated path coefﬁcients are represented in Figures 1 and 2 and Tables 4 and 5. Models 1 and 2 are similar, and they only differ in the ﬁnal section of the diagrams. 108 J. PASCUAL ET AL. TABLE 3. Intercorrelations among variables assumed to determine number of offspring M M AAM MBI AMFB AMLB CP CONTRAC YFM P ABOR NLB AAM MBI AMFB AMLB CP CONTRAC YFM P ABOR NLB ⫺0.109 ⫺0.084 ⫺0.034 0.131 0.766 0.028 ⫺0.038 0.452 0.499 0.528 ⫺0.155 ⫺0.214 0.516 ⫺0.348 0.628 0.048 ⫺0.255 0.043 ⫺0.314 ⫺0.277 0.018 0.219 0.250 0.162 0.210 0.038 0.255 0.343 0.014 ⫺0.251 ⫺0.237 ⫺0.310 0.214 ⫺0.471 ⫺0.085 0.150 ⫺0.061 ⫺0.035 0.043 0.030 ⫺0.024 ⫺0.080 ⫺0.113 0.079 0.383 ⫺0.006 ⫺0.245 ⫺0.289 ⫺0.347 0.243 0.544 ⫺0.062 0.164 0.923 0.045 TABLE 4. Model 1: degrees of determination by variables of reproductive history of women Determination AMFB Age at menarche Age at marriage AMLB AMFB MBI CP Contraceptives AMLB Pregnancies Age at marriage CP Fig. 1. Model 1. Path diagram shows suggested interconnections and path coefﬁcients among factors. Determination Path models of fertility The ﬁrst event of the childbearing period of a woman (i.e., age of mother at ﬁrst birth; AMFB) is almost completely determined by two parameters of a different nature: age at menarche (a biological trait), and age at marriage (a sociocultural trait). Nevertheless, age at marriage is clearly the most explanatory factor, because it explains 76.6% of the variability, while age at menarche explains only 5.1% of the variability. This may be related to the late mean age at marriage of these women, which moves the biological event that marks the beginning Combined Total Unexplained 0.051 0.766 ⫺0.022 0.795 0.205 0.264 0.235 0.014 0.513 0.487 0.043 0.471 ⫺0.079 0.435 0.565 0.138 0.331 ⫺0.092 0.377 0.623 TABLE 5. Model 2: degrees of determination by variables of reproductive history of women AMFB Age at menarche Age at marriage AMLB AMFB MBI CP Contraceptives AMLB Pregnancies Age at marriage YFM CP Fig. 2. Model 2. Path diagram shows suggested interconnections and path coefﬁcients among factors. Direct Direct Combined Total Unexplained 0.051 0.766 ⫺0.043 0.774 0.226 0.264 0.235 0.014 0.513 0.487 0.043 0.471 ⫺0.079 0.435 0.565 0.107 0.042 0.352 ⫺0.111 0.389 0.611 of fertile life away from the next biological event of the reproductive history of a woman (age at ﬁrst birth, that marks the beginning of childbearing period). Only 20.5% of the variability of AMFB is unexplained by these two factors (see Tables 4 and 5). Two cuasi biological variables, maternal age at ﬁrst birth and mean birth interval, explain together more than 50% of the variability of the age of a mother at last birth (both menarcheal and nuptial ages explain little additional variability). The beginning of the childbearing period of a woman and the rhythm of conception of the offspring (mainly determined by breastfeeding period, the use of contraceptives, and either temporal or deﬁnitive spousal separation) determine maternal age at last birth. Path coefﬁcients are very similar to correlation coefﬁcients, which show that these two factors act in an independent path. In a general way, a woman would 109 FERTILITY IN TIERRA DEL FUEGO have last birth at a late age if she had her ﬁrst birth at a late age, and her children are separated by long birth intervals. In the same way, a woman would ﬁnish her childbearing period at early age if she had her ﬁrst birth at an early age and her mean birth interval was short. The residual value indicates the importance of unconsidered determinants in the variability of maternal age at last birth. The length of the childbearing period is mainly determined by the age of the mother at last birth and the use of contraceptives, in such a way that the higher the age of a mother or the use of contraceptives, the longer the childbearing period. The length of childbearing period depends more on the endpoint of the interval (age of mother at last birth explains 47.1% of the variability) than the beginning of the span (age of mother at ﬁrst birth could explain directly only 12.11%). A path diagram that includes simultaneously both a direct effect of AMFB and a direct effect of AMLB would be inconsistent due to a low multicollinearity among variables; this is because the length of the childbearing period is not a natural event of women, but is a variable deﬁned as the difference between AMLB and AMFB. Contraception is an important determinant of the length of the childbearing period. Thus its use could be not only a stopping strategy, but also a spacing control action. Determinants of the number of pregnancies are different in both models presented here. The age of marriage and the length of the childbearing period explain 37.7% of the variability of the number of pregnancies in model 1. The negative sign of path coefﬁcients indicates an opposite connection between variables. In general, women who married at late ages and who enjoyed a long childbearing period would have a number of live births lower than those women who married early and who had a short childbearing period. In this sense, a long separation between ﬁrst and last birth does not provide a greater number of children. In other words, the beginning of married life, the rhythm of conception, and maternal age at last birth are the main determinants of the number of pregnancies of Tierra del Fuego’s women. The age of marriage, the fertile years during marriage, and the length of the childbearing period are the main determinants of fertility in model 2. The three variables explain 38.9% of the variability of the number of pregnancies. Differences between models are low. Model 1 considers the event that deﬁnes the beginning of a stable connection (age at marriage), whereas model 2 adds the duration of marriage (considering deﬁnitive separation of spouses, divorce, widowhood, repeated marriages, or deﬁnitive loss of fecundity by surgery). Moreover, the sterile age for women (age when women experience a natural or an artiﬁcial menopause) was added as the most explanatory factor of the variability of fertile years during marriage. In this sense, the endpoint (and not the beginning of TABLE 6. Analysis of variance showing influence of age at marriage in number of live births1 Age at marriage ⬍20 years 20–24 years 25–29 years ⱖ30 years Mean of NLB SD N 3.89 3.48 3.37 2.38 1.96 1.89 1.46 1.0 56 52 35 16 Difference between means is signiﬁcant (P ⬍ 0.05) between women married before 25 years old and women married after 30 years old. 1 the marriage) explains the variability of the fertile years during the marriage. Differences in both age of marriage and birth interval are the most explanatory factors of the sterile age for women who experience an artiﬁcial menopause or sterilization (explaining 73.4% of variability), but these factors only explain roughly 24% of the sterile age of women with natural menopause. Finally, the number of pregnancies of a woman is obviously the main determinant of her number of live births, because it explains 85.2% of the variability. Abortions and stillbirths explain the residue. Additional results Several additional analyses were included to complete the general view of the inﬂuence of some factors in the fertility of the population. First, an analysis of variance that shows the inﬂuence of age at marriage in the number of live births is given in Table 6 (F ⫽ 3.145; df, 3,155; P ⫽ 0.027). Women with late age at marriage have fewer live births than women who get married early. On the other hand, women who live with less than 10 years of fertile marriage (3.28 live births, SD 1.20, n ⫽ 32), women who live from 10 –19 years (3.43 live births, SD 1.57, n ⫽ 30), and women who live with more than 19 years of fertile marriage (3.82 live births, SD 2.14, n ⫽ 44) do not show statistical differences in their fertility (F ⫽ 0.970; df, 2,103; P ⫽ 0.383). Finally, Table 7 shows no signiﬁcant differences (F ⫽ 0.754; df, 2,116; P ⫽ 0.473) in age of the mother at last birth according to deﬁnitive loss of fecundity of women. There are no signiﬁcant differences (F ⫽ 0.700; df, 2,125; P ⫽ 0.498) in mean number of live births by reason for becoming infertile. Finally, the effect of contraception in fertility of women from Tierra del Fuego is pointed out with two simple comparisons. First, birth intervals where women use contraception show a mean length of 4.068 years (SD 3.439, n ⫽ 94), and birth intervals where women do not use contraception show a mean length of 2.099 years (SD 1.416, n ⫽ 191). The differences are statistically signiﬁcant (t ⫽ 5.335; df, 283; P ⫽ 0.000). On the other hand, the fertility of women who afﬁrm the use of contraceptives (X ⫽ 3.33, var ⫽ 1.85, n ⫽ 114) is not signiﬁcantly different (t ⫽ 0.614; df, 154; P ⫽ 0.540) from the fertility 110 J. PASCUAL ET AL. TABLE 7. Mean age of mother at last birth and mean number of live births according to reason for definitive end of fecundity Age at last birth SD N NLB SD N 32.55 32.31 30.90 6.84 6.45 4.44 66 22 31 3.27 3.09 3.65 2.11 1.31 1.38 74 23 31 Natural menopause Menopause due to surgery Sterilization of women who do not use them (X ⫽ 3.60, var ⫽ 7.42, n ⫽ 42). DISCUSSION The recent origin of the colonizing population of the Chilean part of Tierra del Fuego, together with the socioeconomic characteristics of this austral population (semirural economy, and general (female) low level of education), lead us to think that the studied population keeps similar reproductive patterns from other colonizing populations or developing societies (such as early age at marriage, a long childbearing period, or high fertility) (Bongaarts and Potter, 1983; Cleland et al., 1984; Wood, 1994; Castro Martı́n and Juarez, 1995). Instead of this preliminary assumption, results conﬁrm that the population of Tierra del Fuego has experienced a swift modernization of these reproductive patterns throughout the 20th century, which place the population in an advanced phase of the fertility transition (Bongaarts and Potter, 1983). The age at menarche in human populations is described as between 11–15 years, with maximum values of 17 years in poor nutritional populations (Livi-Bacci, 1993; Thomas et al., 2001). Early ages at menarche are more common in developed populations, where both nutritional and sanitary conditions are better. The age at menarche of Tierra del Fuego women is one of the latest ages described in South America. South American populations of European origin show earlier ages at menarche than native populations, due to the genetic pool or better socioeconomic development (Pascual et al., 2003; Garcı́a-Moro et al., 2000; Greksa, 1990; DankerHopfe, 1986). In contrast, human populations show mean ages at natural menopause between 42–54 years (Leridon and Menken, 1979; Leidy, 2001). Mean age at natural menopause of women from Tierra del Fuego is a central value inside the range. The average length of fertile span was described in human populations as between 26.8 years in Agta women from the Philippines (Goodman et al., 1985) and 35.9 years in a sample of Minnesota women (Treloar, 1974). Women with natural menopause from Tierra del Fuego show a long fertile span (34.12 years). Nevertheless, an important proportion of women (32.97%) reduced their fertile span by 36.08% due to surgery, with an expected strong consequence in their ﬁnal fertility. However, this repercussion is not observed in the population, as we show later. Age at marriage is considered one of the main determinants of fertility, especially in natural fertility populations (van de Walle, 1978; Varea, 1993). Age at marriage is a signiﬁcant determinant of the duration of marriage, and a close connection between duration of marriage and fertility is frequently described in human populations (Busﬁeld, 1972; Bumpass and Mburugu, 1977; Bongaarts and Potter, 1983; Aryal, 1991). This connection is not signiﬁcant in Tierra del Fuego. When this connection exists, it becomes an indirect sign of nonexistent or infrequent contraceptive practices, in such a way that women have children during all their reproductive years (Loebner and Driver, 1973). In fact, duration of marriage can be considered an indirect measure of the exposure to risk of pregnancy, in populations where illegitimate children are scarce (Josephson, 2002). On the other hand, maternal age at marriage is also important in these populations, because a woman’s reproductive capacity is greatest in her late teens and early twenties. Age at marriage is an important determinant of fertility if it is not late and the time lag between wedding and cohabitation is short (Loebner and Driver, 1973). The age at marriage of women from Tierra del Fuego is related to their ﬁnal number of live births (Table 5), but this connection is not seen across the duration of marriage. The comparison of total fertility rate (3.3) and total marital fertility rate (6.4) shows that the speciﬁc features of marriage are one of the main determinants of fertility in Tierra del Fuego. The late mean age at marriage and the frequent breakup of marriage reduce the potential fertility. Age of the mother at ﬁrst birth is mainly determined by sociocultural regulations in the majority of developed societies and other rural populations like Aitourir, Morocco (Salvat and Crognier, 1993), or the Mapuche population of Rı́o Grande, Argentine (Crognier et al., 1996). Menarcheal age or high levels of fecundability were related to the beginning of the childbearing period as well (Komlos, 1989; Udry and Cliquet, 1982). The age at marriage is the main factor which determines the maternal age at ﬁrst birth in women from Tierra del Fuego. On the other hand, most of the childbearing periods in human populations have been described as between the 5.2 years of Goessel Mennonites (USA) (Stevenson et al., 1989) and the 16.64 years of the Amizmiz population (Marrakech, Morocco) (Varea, 1990). Among South American populations, lengths of childbearing period of 13 years, 13.1 years, and 14.95 years were described, respectively, in the Shipibo Peruvian Amazon (Hern, 1994), the Aymara peasant society of the Bolivian Altiplano (Crognier et al., 2002), and Mapuche women of the province of Rı́o Negro (Argentine) (Crognier et al., 1996). Women from Tierra del Fuego show a relatively FERTILITY IN TIERRA DEL FUEGO short childbearing period if the length of the interval is compared with the long fertile span. In fact, women from Tierra del Fuego only use 26.05% of their fertile span, to reach 3.3 live births per woman. Available data in other human populations show that women from Maragateria (Spanish rural region; Bernis, 1990) use 30% to bear 4.4 live births, and women from Amizmiz (Morocco; Varea, 1993) use 59% to have 8 live births. The mean length of the childbearing period is closely related to both long birth intervals and low fertility. The relatively widespread use of contraceptives in the Chilean population of Tierra del Fuego was unknown before this study. Contraception is widely extended in developed populations (Frejka and Ross, 2001), but this is not the case in developing societies, as studied elsewhere (Fort, 1989; Parker and Ross, 1992). Their use was described both as a stopping behavior (to limit the size of progeny) and as a spacing behavior (to separate births). The study of the age of the mother at last birth provides information about the stopping behavior of the population. The mean age at last birth is remarkably invariant in natural fertility populations. With few exceptions, the means fall in the 39 – 41year range (Bongaarts and Potter, 1983; Desjardins et al., 1994). These results suggest that the timing of the onset of natural sterility varies little among populations, except in cases where sexual infectious diseases are widespread. Then, factors such as taboos against intercourse at older ages (Leidy, 1993), disease-induced sterility (Pennington and Harpending, 1991; Jenkins, 1993), or the use of contraception and sterilizations (Leidy, 1999; Barroso et al., 2003) can vary the values in human populations. In Tierra del Fuego, women who were sterilized showed an earlier age at last birth than natural menopausal women, but the differences were not statistically signiﬁcant (Table 6). Nevertheless, the early mean age of a mother at last birth (even in natural menopausal women) suggests that the use of contraception in Tierra del Fuego limits the number of live births. On the other hand, with regard to spacing behavior, the length of birth intervals is longer (nearly 2 years) for women who use contraceptives than for women who do not use them. The results also suggest the use of contraception to space births. The fertility of women who afﬁrm the use of contraceptives is not signiﬁcantly different from the fertility of women who do not use them. Nevertheless, contraception is infrequent in women with a high number of live births, and women who have used contraceptives in their reproductive lives mainly reach a number of three or four live births. This progeny size is also the most represented in sterilized women. At the same time, natural menopausal women do not have a greater number of live births than sterilized women. These results are indirect evidence of the use of contraceptives and of sterilization to regulate women’s fertility to reach their desired family size. In conclusion, contracep- 111 tion has become an important determinant of fertility in the population, especially inﬂuencing both the rhythm and the endpoint of the childbearing period. In conclusion, the fertility model of Tierra del Fuego moves away from described models of traditional natural fertility populations (such as the Amish or Mennonites; Felt et al., 1990; Hewner, 1998; Greksa, 2002) or European historical populations (Henry, 1961; Knodel and Wilson, 1981). In that sense, comparisons with other colonizing populations (Campbell and Wood, 1988) show that rural Canadians during the early eighteenth century (Charbonneau et al., 1987), the Hutterites during the late nineteenth and early twentieth centuries (Eaton and Mayer, 1953), the Mormons during their phase of westward expansion in the nineteenth century (Mineau et al., 1979), and the colonizing population of California (Garcı́a-Moro et al., 1997b) share the main fertility determinants described for those natural fertility populations. The age at marriage (the beginning of the marital life) and marriage duration are the main determinants of fertility in this kind of population. The dissolution of a marriage before the end of the reproductive period and the absence of contraception are also considered important determinants of fertility. On the other hand, among the closest geographical populations, the reproductive patterns of the rural Mapuche population of Rı́o Negro (Patagonian Argentine) offer the only fertility model known (Crognier et al., 1996). The use of contraception is widespread in the population, but is mainly limited to reproductive years after couples reach the desired number of children. Furthermore, there is no regulation of fertility access, because marriage is not universal in the population. Finally, path analysis was applied in a few studies of human fertility. Woman’s fertility is largely a direct function of both age at marriage and duration of marriage in populations from Central India (Loebner and Driver, 1973); in contrast, the birth interval and the span of reproductive life (deﬁned as the interval between age at marriage and age at last birth) are the main determinants of fertility in an Amizmiz Berber population (Crognier, 1996). Women from the Chilean part of Tierra del Fuego show a different pattern from natural fertility populations. The duration of marriage was considered by means of fertile years of marriage, and this factor only explains 4.2% of the variability in the number of pregnancies. This result suggests an extensive and effective use of contraceptives that reduces the importance of duration of marriage. On the other hand, the beginning of the marital union is a significant determinant of fertility in many of the cited populations (including Tierra del Fuego), because marriage is almost universal and fertility is mostly limited to these unions. An example of the nonexistence of this connection is the Mapuche population of the Patagonian Argentine. Finally, Tierra del Fuego and the Amizmiz populations share some reproduc- 112 J. PASCUAL ET AL. tive characteristics. Fertility analyses of both populations reveal the importance of the beginning of the conjugal union, the rhythm of conception, and the end of the childbearing period. CONCLUSIONS In summary, the method of path analysis applied to analyze the fertility determinants shows a fertility pattern deﬁned by the importance of biological and sociocultural factors. A late age of women at marriage, the rhythm of conception, and practices of contraception limiting and spacing births are the main determinants of fertility in the Chilean part of Tierra del Fuego. Our next efforts will study in depth 1) some aspects of the reproductive histories of women from Tierra del Fuego, and 2) the quantitative importance of fertility determinants in the population. LITERATURE CITED Aryal RH. 1991. Socioeconomic and cultural differentials in age at marriage and the effect on fertility in Nepal. J Biosoc Sci 23:167–178. Barroso A, Bernis C, Arias S, González MB, Montero P, Varea C. 2003. Aspectos antropológicos y epidemiológicos sobre la histerectomı́a y la oforectomı́a en mujeres españolas. In: Aluja MP, Malgosa A, Nogués RM, editors. Antropologı́a y biodiversidad. Barcelona: Edicions Bellaterra S.L. p 513–526. Bergsten-Brucefors A. 1976. A note on the accuracy of recalled age at menarche. Ann Hum Biol 3:71–73. Bernis C. 1990. Canvis en els models de reproducció humana: fets i problemes teórics. In: Poblacions, societats i entorns. Barcelona: Barcanova. p 260 –281. Bongaarts J, Potter R. 1983. Fertility, biology and behavior. An analysis of the proximate determinants. New York: Academic Press. Bumpass LL, Mburugu EK. 1977. Age at marriage and completed family size. Soc Biol 24:31–37. Busﬁeld J. 1972. Age at marriage and family size: social causation and social selection hypotheses. J Biosoc Sci 4:117–134. Campbell KL, Wood JW. 1988. Fertility in traditional societies. In: Diggory P, Potts M, Teper S, editors. Natural human fertility. Social and biological determinants. London: MacMillan Press and the Eugenics Society. p 39 – 69. Castro Martı́n T, Juarez F. 1995. The impact of women’s education on fertility in Latin America: searching for explanations. Int Fam Plan Perspect 21:52–57. Charbonneau H, Desjardins B, Guillemette A, Landry Y, Legare J, Nault F. 1987. Naissance d’une population. Les Français établis au Canada au XVIIe siècle. I.N.E.D., Travaux et Documents. Paris et Montréal. P.U.F. et Presses de l’Université de Montreal. Clay DC, Vander Haar JE. 1993. Patterns of intergenerational support and childbearing in the Third World. Popul Stud 47: 67– 83. Cleland J, Casterline JB, Singh S, Ashurst H. 1984. The effects of nuptiality, contraception and breastfeeding on fertility in developing countries. Int Fam Plan Perspect 10:86 –92. Crognier E. 1996. Behavioral and environmental determinants of reproductive success in traditional Moroccan Berber groups. Am J Phys Anthropol 100:181–190. Crognier E, Elizondo S, Caratini AL, Zubieta NI, Niborsky R, Carnese RF. 1996. Los comportamientos reproductivos de la población mapuche de la provincia de Rı́o Negro. Rev Argent Antropol Biol 1:86 –97. Crognier E, Villena M, Vargas E. 2002. Helping patterns and reproductive success in Aymara communities. Am J Hum Biol 14:372–379. Danker-Hopfe H. 1986. Menarcheal age in Europe. Yrbk Phys Anthropol 29:81–112. Desjardins B, Bideau A, Brunet G. 1994. Age of mother at last birth in two historical populations. J Biosoc Sci 26:509 –516. Eaton JW, Mayer AJ. 1953. The social biology of very high fertility among the Hutterites. The demography of a unique population. Hum Biol 25:206 –264. Felt JC, Ridley JC, Allen G, Redekop C. 1990. High fertility of Old Colony Mennonites in Mexico. Hum Biol 62:689 –700. Fort AL. 1989. Investigating the social context of fertility and family planning: a qualitative study in Peru. Int Fam Plan Perspect 15:88 –95. Frejka T, Ross JA. 2001. Paths to subreplacement fertility: the empirical evidence. In: Global fertility transition. New York: Population Council. p 213–254. Garcı́a-Moro C, Hernández M, Lalueza C. 1997a. Estimation of the optimum density of the Selk’nam from Tierra del Fuego: inferences about human dynamics in extreme environments. Am J Hum Biol 9:699 –708. Garcı́a-Moro C, Toja DI, Walker PL. 1997b. Marriage patterns of California’s early Spanish-Mexican colonists (1742–1876). J Biosoc Sci 29:205–217. Garcı́a-Moro C, Hernández M, Moral P, Gonzalez-Martin A. 2000. Epidemiological transition in Easter Island (1914 –1996). Am J Hum Biol 12:371–381. Gober P. 1997. The role of access in explaining state abortion rates. Soc Sci Med 44:1003–1016. Goodman MJ, Estioko-Grifﬁn A, Bion Grifﬁn P, Grove JS. 1985. Menarche, pregnancy, birth spacing and menopause among the Agta women foragers of Cagayan Province, Luzon, the Philippines. Ann Hum Biol 12:169 –177. Greksa LP. 1990. Age of menarche in Bolivian girls of European and Aymara ancestry. Ann Hum Biol 17:49 –53. Greksa LP. 2002. Population growth and fertility patterns in an Old Order Amish settlement. Ann Hum Biol 29:192–201. Henry L. 1961. Some data on natural fertility. Eugen Q 81–91. Hern WM. 1994. Cultural change, polygyny, and fertility among the Shipibo of the Peruvian Amazon. South Am Indian Stud 4:77– 86. Hewner SJ. 1998. Fertility, migration, and mortality in an Old Order Amish community. Am J Hum Biol 10:619 – 628. Jenkins CL. 1993. Fertility and infertility in Papua New Guinea. Am J Hum Biol 5:575–583. Josephson SC. 2002. Does polygyny reduce fertility? Am J Hum Biol 14:222–232. Knodel J, Wilson C. 1981. The secular increase in fecundity in German village populations: an analysis of reproductive histories of couples married (1750 –1899). Popul Stud 35:53– 84. Komlos J. 1989. The age at menarche and age at ﬁrst birth in an undernourished population. Ann Hum Biol 16:463– 466. Leidy LE. 1993. The practice of terminal abstinence in Nigeria and Cameroon. Am J Hum Biol 5:565–573. Leidy LE. 1999. Effect of exclusion: rates of hysterectomy and comparisons of age at natural menopause. Am J Hum Biol 11:687– 693. Leidy LE. 2001. Menopause as a measure of population health: an overview. Am J Hum Biol 13:429 – 433. Leridon H, Menken J, editors. 1979. Natural fertility. Patterns and determinants of natural fertility: proceedings of a seminar on natural fertility. Liège: Ordina Editions. Li CC. 1975. Path analysis. A primer. Paciﬁc Grove, CA: Boxwood Press. Livi-Bacci M. 1993. Introducción a la demografı́a. Barcelona: Editorial Ariel S.A. Loebner H, Driver ED. 1973. Differential fertility in Central India: a path analysis. Am J Phys Anthropol 10:329 –350. Madrigal L. 1991. The reliability of recalled estimates of menarcheal age in a sample of older women. Am J Hum Biol 3:105– 110. Madrigal L, Relethford JH, Crawford MH. 2003. Heritability and anthropometric inﬂuences on human fertility. Am J Hum Biol 15:16 –22. Mineau GP, Bean LL, Skolnick M. 1979. Mormon demographic history II: the family life cycle and natural fertility. Popul Stud 33:429 – 446. FERTILITY IN TIERRA DEL FUEGO Parker W, Ross JA. 1992. Contraceptive use and commodity costs in developing countries, 1990 –2000. Int Fam Plan Perspect 18:4 –9. Pascual J, Garcı́a-Moro C, Hernández M. 2000. Non-seasonality of births in Tierra del Fuego (Chile). Ann Hum Biol 27:517–524. Pascual J, Garcı́a-Moro C, Hernández M. 2003. Intervalo fértil en mujeres colonizadoras de Tierra del Fuego (Chile). In: Aluja MP, Malgosa A, Nogués RM, editors. Antropologı́a y biodiversidad. Barcelona: Edicions Bellaterra S.L. p 293–239. Pennington R, Harpending H. 1991. Infertility in Herero pastoralists of southern Africa. Am J Hum Biol 3:135–153. Salces I, Rebato EM, Susanne C, San Martı́n L, Rosique J. 2001. Familial resemblance for the age at menarche in Basque population. Ann Hum Biol 28:143–156. Salvat M, Crognier E. 1993. Patrones reproductores en la población de Aitourir (Marruecos). In: Bernis C, Varea C, Robles F, González A, editors. Biologia de poblaciones humanas: problemas metodológicos e interpretación ecológica. Ediciones de la Universidad Autónoma de Madrid. p 387–398. Shen C, Williamson JB. 1999. Maternal mortality, women’s status, and economic dependency in less developed countries: a cross-national analysis. Soc Sci Med 49:197–214. Stevenson JC, Everson PM, Crawford MH. 1989. Changes in completed family size and reproductive span in Anabaptist populations. Hum Biol 61:99 –115. 113 Thomas F, Renaud F, Beneﬁce E, De Meeüs T, Guegan JF. 2001. International variability of ages at menarche and menopause: patterns and main determinants. Hum Biol 73:271–290. Treloar AE. 1974. Menarche, menopause, and intervening fecundability. Hum Biol 46:89 –107. Tryggvadóttir L, Tulinius H, Larusdóttir M. 1994. A decline and a half in mean age at menarche in Iceland. Ann Hum Biol 21:179 –186. Udry JR, Cliquet RL. 1982. A cross-cultural examination of the relationship between ages at menarche, marriage, and ﬁrst birth. Demography 19:53– 63. van de Walle E. 1978. La nupcialidad y fecundidad marital. In: Glass DV, Revelle R, editors. Población y cambio social: estudios de demografı́a histórica. Madrid: Ed Tecnos. p 143–156. Varea C. 1990. Patrones reproductores y fertilidad en una población tradicional de Marruecos (Amizmiz, Marrakech). Tesis doctoral. Universidad Autónoma de Madrid. Varea C. 1993. Marriage, age at last birth and fertility in a traditional moroccan population. J Biosoc Sci 25:1–15. Westoff CF. 1990. Reproductive intentions and fertility rates. Int Fam Plan Perspect 16:84 – 89. Wood JW.1994. Dynamics of human reproduction. Biology, biometry, demography. New York: Aldine de Gruyter.