Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://ecology.genebee.msu.ru/3_SOTR/CV_Terekhin_publ/2004_Birthweight_JEvolBio.pdf Äàòà èçìåíåíèÿ: Mon Mar 16 11:01:21 2009 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 22:02:22 2012 Êîäèðîâêà:

JEB
Journal Name

7

0

5

Manuscript No.

B

Dispatch: 30.1.04 Author Received:

Journal: JEB CE: Kumar No. of pages: 12 PE: Sri

doi:10.1111/j.1420-9101.2004.00705.x

Human birthweight evolution across contrasting environments
F. T H OMAS , * A . T . T ER IOKH IN ,* E. V. BUDILOVA, S. P. BROWN, Þ F. R E NAUD* & J. F. GUEGAN*

` *Centre d`Etude sur le Polymorphisme des Micro-Organismes CEPM/UMR CNRS-IRD 9926 Equipe: `Evolution des Systemes Symbiotiques', IRD, Montpellier Cedex, France Department of Biology, Moscow State University, Moscow, Russia ´ ` ÞInstitut des Sciences de l'Evolution, Universite de Montpellier II, Place Eugene Bataillon, Montpellier Cedex, France

Keywords:
birthweight; chronic diseases; Homo sapiens; individual-based agent modelling; infectious diseases.

Abstract

dominate (Bogin & Smith, 2000), prenatal growth and birthweight in humans are unlikely to be solely at the mercy of external constraints. Several studies indicated Human populations are characterized by important that size and weight at birth in humans are heritable (e.g. differences in birthweight (Kleinman & Kessel, 1987; Cramer, 1995; Le et al., 1996; Shiono et al., 1997; 1,2 Langhoff-Roos et al., 1987; Emanuel, 1992; Towne et al., Bereczkei et al., 2000; Bogin & Smith, 2000; Stinson, 3 1993; Klebanoff et al., 1998). From an evolutionary point of view, human mothers, like other mammals, are 2000; Vangen et al., 2002). Causes for this variation are expected to maximize their lifetime reproductive success potentially numerous as many variables have been by delivering infants with an optimal birthweight given shown to influence prenatal growth and birthweight in extant circumstances (Trivers, 1974; Blurton Jones, humans (e.g. maternal energy supply, maternal stature, 1978; Bereczkei, 1993; Bereczkei et al., 2000). Typically, physical work, stress, temperature, disease status, smothe relationship of birthweight to infant mortality is king status, gestation length and altitude; see Abell et al., U-shaped as a result of stabilizing selection (Futuyma, 1991; Brabin, 1991; Emanuel et al., 1992; Zhang & 1998; Stearns & Hoekstra, 2002). However, little is Savitz, 1992; Defo & Partin, 1993; Koupilova et al., known concerning the short- and long-term fitness 2000; Wells, 2002). Although public health workers consequences of birthweight variation in the different traditionally assumed that nongenetic factors prekinds of environments inhabited by human populations ´´ Correspondence: Frederic Thomas, Centre d'Etude sur le Polymorphisme across the world (Bereczkei et al., 2000; Lummaa & des Micro-Organismes, CEPM/UMR CNRS-IRD 9926 Equipe: `Evolution Clutton-Brock, 2002). ` des Systemes Symbiotiques' IRD, 911 Avenue Agropolis, B.P. 64501 According to evolutionary theory, intra- and inter34394 Montpellier Cedex 5, France. specific differences in life history traits result for a large Tel.: (33) 04 6741 6318; fax: (33) 04 6741 6299; part from differences in the characteristics of the e-mail: fthomas@mpl.ird.fr

Introduction

UN CO RR EC TE D

We explore from both theoretical and empirical perspectives the hypothesis that a significant part of the worldwide variability in human birthweight results from adaptive responses to local selective pressures. We first developed an agent-based model to simulate the process of evolutionary selection on life history strategy, and then we performed a comparative analysis across 89 countries worldwide. The model illustrates that optimal birthweight depends on which fitness-reducing risk locally predominates (somatic diseases, parasitic diseases or adverse environmental conditions). When fitness variations between individuals mainly result from somatic diseases (e.g. industrialized countries), or conversely from infectious and parasitic diseases (e.g. developing countries), selection is expected to favour individuals producing larger children. Conversely, when environmental risks increase in relative importance, selective pressures for producing children with high birthweight are reduced. The comparative analysis supports these theoretical expectations, in particular the finding that birthweight is higher than predicted in highly parasitized countries.

J . EVOL. BI OL. ª 20 04 BL ACK WELL P UBL I S H ING L TD

PR OO F

1


2

F. THOM AS ET AL .

UN CO RR EC TE D

environment (Roff, 1992; Stearns, 1992). Even if we are far from fully understanding the selective pressures experienced by human populations during the course of evolution, it is safe to consider that anywhere on earth, humans faced (and still face nowadays) three potential risks of major fitness reduction: individuals can die or be severely impaired because of (i) chronic somatic diseases (e.g. cancer, cardiovascular disease), (ii) infectious and parasitic diseases and/or (iii) adverse environmental conditions (e.g. famine, dryness, accidents, disturbances). Health statistics clearly show that these risks strongly vary between human populations in relation to geographic, cultural and economic factors. Interestingly, different predictions can be made on how natural selection should adjust birthweight according to which risk of fitness-reduction predominates. In stable, well-resourced, low-parasite environments (i.e. most modern industrialized countries), somatic diseases are likely to be an important source of fitness variation between individuals. Because infants with a low birthweight are at a higher risk of expressing chronic diseases later in life (e.g. cardiovascular diseases, diabetes, certain cancers, impairment of hearing and vision; see for instance Barker et al., 1989; Kopp, 1990; Robertson et al., 1990; Godfrey & Barker, 2000; Cicognani et al., 2002; Falkner, 2002; Nafstad et al., 2002; Tulassay & Vasarhelyi, 2002), selection in these environments should presumably favour individuals producing larger children (prediction 1). Even if some of these somatic diseases occur late in life (i.e. after reproduction), they are likely to be detrimental to an individual's fitness as they reduce its capacity to deliver grandparental care (inclusive fitness). A second prediction is that in countries where the risks of parasitic infections are extreme (i.e. numerous developing countries), we also expect, all things being equal, women to deliver infants with a high birthweight (prediction 2). Indeed, infants with low birthweight generally have an increased vulnerability to infectious diseases because of impaired immune function (Ferguson, 1978; Victora et al., 1988, 1989, 1990; Cerqueiro et al., 1990; Bukenya et al., 1991; Ittiravivongs et al., 1991; Fonseca et al., 1996; Chandra, 1997, 1999; Moore et al., 1999). Given that offspring mortality (because of infections), more than fertility, is likely to be the primary determinant of fitness variation between reproducing females, human mothers in parasite-rich environments will have a particular reproductive interest in producing larger, more resistant, children. Finally, in environments where adverse environmental conditions predominates (e.g. famine, dryness, accidents, disturbances...), selective pressures for producing large offspring are likely to be relaxed because the negative impacts of environmental factors on individual fitness are largely birthweight-independent. Thus, because in these cases the fitness costs incurred by the maternal organism when producing large children (e.g. reduced survival, lower probability of subsequent reproduction, see Ber-

Materials and methods
Evolutionary modelling

We used an agent-based model to simulate the process of evolutionary selection on life history strategy (see Mangel, 1990). The choice of an agent-based model instead of using more computationally advantageous dynamic optimization models (Mangel & Clark, 1988; 4 Perrin & Sibly, 1993; McNamara & Houston, 1996; Teriokhin, 1998; Kozlowski & Teriokhin, 1999) was made mainly because of the need to optimize birthweight alone. In optimization models, birthweight cannot be optimized directly as it enters into dynamic equations as an initial value (Mangel et al., 1994). Another limitation of optimization modelling, which can be overcome by using agent-based evolutionary models, is that it requires an explicit expression for an optimality criterion. Different optimality criteria (lifetime reproductive success, intrinsic rate of increase etc.) may be adequate for a given modelled situation, depending on the way in which the environment influences individual survival and fertility (Mylius & Diekmann, 1995; Teriokhin, 2002). Thus, the choice of a suitable criterion may be a nontrivial task. On the contrary, the optimality requirement in agent-based models is included implicitly and naturally as the ultimate dominance of a genotype in a population as a result of its competition with other genotypes. The idea of agent-based modelling consists in simulating iteratively, passing from time step t to t + 1, the evolution of a population of individuals (agents), each being eventually able to die or to give birth to new individuals. At any time step t, each individual of the population is characterized by a set of state variables (such as age or body mass), which can change with time, and by its genotype, which cannot change during the individual's lifetime. The descendent inherits the genotype of its parents modified by possible mutations. The environmental setting of a population is characterized by a set of time-independent state-variables. Thus, we assumed that evolution occurs in a fixed environment.

J. EVO L. BIO L . d o i :10 . 1 111 /j.142 0-910 1.20 04.00 705 . x ª 20 04 B L A C K WELL P UB LIS H ING L TD

PR OO F

eczkei et al., 2000) are less well compensated by reproductive advantages, natural selection should favour individuals producing smaller babies (prediction 3). The aim of this paper was to evaluate both from a theoretical and an empirical perspective the hypothesis that human birthweight displays regional variation as a consequence of evolutionary adaptation of populations to their environment. We first used an agent-based model to simulate the process of evolutionary selection for the best life history strategy in contrasting environments. Then, we performed a multivariate comparative analysis in which we examined more specifically the role of disease-causing agents on birthweight evolution across 89 countries from around the world.


Human birthweight evolution

3

The dynamic equation for body weight is

where Ei(t) is the amount of disposable energy which the individual may spend either on growth or reproduction depending on the value of the single decision variable ui(t). This variable may take one of only two values: one or zero. If ui(t) ¼ 1 all the energy is allocated to growth and if ui(t) ¼ 0 all the energy is allocated to reproduction. Thus, in the latter case growth is replaced by giving birth to a child j with birthweight Bj equalling Bj ¼ ½1 þ ui Ït÷Ei Ït÷

The value of ui(t), in turn, depends on the value of the state variable Wi(t): if Wi(t) is less than some threshold value Gi then ui(t) ¼ 1, otherwise ui(t) ¼ 0. So, we can say that the control system of an individual in our model consists of a single two-state neurone characterized by its threshold Gi which may be also considered as the individual's genotype. We used the commonly accepted allometric dependence of the energy produced by an individual on its weight (West et al., 1999) Ei Ït÷ ¼ aWib Ït÷

In our simulations we set the parameter b to 0.5 which ´ was estimated using real human data (see Guegan et al., 2000). Our experience shows that this value proves more adequate to models of human life history than the value b ¼ 0.75 postulated by Kleiber's law (West et al., 1999).

J. EVO L. BIO L . d oi:10 . 1 111 /j. 1 420 -910 1.200 4.007 05.x ª 20 04 BL ACK WELL P UBL I S H ING L TD

UN CO RR EC TE D
Ai Ït × 1÷ ¼ Ai Ït÷× 1 Wi Ït × 1÷ ¼ Wi Ït÷× ui Ït ÷Ei Ït÷

Nevertheless, we could evaluate the impact of the environment on genotype evolution simply by comparing the results of simulations executed for different environments. The simulated evolution of a population is governed by a set of rules (dynamic equations) which expresses the state of an individual at time t + 1 as a function of its state at time t, taking into account its genotype and the state of environment. We assumed that the genotype of an individual defines the parameters of its internal control system, controlling the state-dependent distribution of energy between growth and reproduction. In the course of simulations under given fixed environmental conditions, the most favourable genotypes spread fastest and ultimately a group of very similar genotypes dominates. Under contrasting environmental conditions other genotypes may dominate, so that the characteristics of the optimal genotype depend critically on the environmental conditions experienced during evolution. In our model, the variables characterizing the individual's state are its age, Ai(t), and body weight, Wi(t). The dynamic equation for the age of an individual i is quite simple

where Qi(t) is the overall mortality rate composed of four components Qi Ït÷ ¼ qe × c edAÏt÷ × qs qp × Bi Bi

The first two components do not depend on individual birthweight and are, in fact, two components of the Gompertz­Makeham equation (Gompertz, 1825; Makeham, 1860). More specifically, while qe does not depend on individual age, the second parameter cedA(t) increases exponentially with age. In our model we interpreted qe as the individual-independent (more precisely, birthweight-independent) component of environmental stress (e.g. because of injuries). It was varied from qe ¼ 0 to 0.008 to imitate aggravation of environmental conditions. In addition, in order to imitate the simultaneous aggravation of alimentary conditions, we tied the increase of qe from 0 to 0.008 with a linearly dependent decreasing function with a varying from 0.57 to 0.47. Henceforth, we will refer to qe as to `(birthweightindependent) environmental stress'. The coefficients c and d of the age-dependent component were taken as constant for all individuals and environmental conditions. We set c ¼ 0.000005 and d ¼ 0.1, corresponding to typical values of these parameters estimated from real data (Gavrilov & Gavrilova, 1991). The third component of mortality rate, qs/Bi, was interpreted as somatic malfunctions (somatic diseases) caused by insufficient birthweight. The parameter qs was set equal to 0.015 for all individuals and environmental conditions. In the following sections, qs will be referred to as (birthweight-dependent) somatic stress. The fourth component, qp/Bi, is the most interesting in our model. It represents the negative effect of parasitic stress (i.e. infectious and parasitic diseases) which becomes stronger for insufficient birthweight. We varied qp from qp ¼ 0 to 0.02 to imitate the aggravation of

PR OO F
Pi Ït÷ ¼ eþQ
i

The value of parameter a reflects several aspects. First, a evidently depends on the units chosen for measuring energy. Here, we measured energy in units of mass, as can be seen from the above dynamic equation for body weight. In principle, the value of a may vary for different individuals, however we considered that all individuals were identical. The value of a also reflects the fact that we only model the allocation of energy to growth and reproduction, and thus it is weighted to ignore contributions to other energetic requirements, e.g. maintenance, defence. Finally, we supposed that the value of a depends on the availability of food. Hence, we made simulations for different values of a to mimic differences in food availability across environments. We supposed that at any time step any individual i may possibly die with a probability
Ït ÷


4

F. THOM AS ET AL .

Parameter Coefficient in the equation of energy production Power in the equation of energy production Coefficient in the Gompertz component of mortality Power in the Gompertz component of mortality Environmental birthweight-independent component of mortality Environmental birthweight-dependent component of mortality Somatic birthweight-dependent component of mortality

Lower Symbol value a b c d q q q 0.47 0.5 0.000005 0.1 0 0 0.015

Upper value 0.57 0.5 0.000005 0.1 0.008 0.2 0.015

Step 0.011 ­ ­ ­ 0.001 0.002 ­

Table 1 Assumptions made on the values of the model parameters.

e p s

parasitic conditions. Further, we will refer to qp as (birthweight-dependent) parasite stress. In addition to Qi(t), two further components of mortality, acting only during the first year of life and so reflecting the effect of infant mortality, were taken into account in the model Qi0 ¼ 10qs 10qp × Bi Bi

Fig. 1 Flowchart illustrating the processes underlying the life functioning of an individual. The dashed arrows represent information flows, the fine arrows indicate direct biological influences and the thick arrows denote the energy flows.

UN CO RR EC TE D
Data
Environment

They are similar to the already considered birthweightdependent components of mortality rate qs/Bi and qp/Bi but are magnified 10 times, corresponding to relative magnitudes of newborn and later-life mortalities as estimated from data (Thomas et al., 2000). The ranges of values for a, q0,qe, as well as the value of qs, were chosen expressly so as to provide realistic values for human life history traits (birthweight, adult weight, age at maturity, life expectancy). Table 1 summarizes the assumptions made for the values of the model parameters. The flowchart in Fig. 1 illustrates the processes underlying the life functioning of an individual.

The model results illustrated in Figs 2 and 3 are based on 99 simulation variants corresponding to 99 combinations of values of qp (from 0 to 0.02 with step 0.002) and qe (from 0 to 0.008 with a step of 0.001). For each simulation, 25 000 iterations were carried out for each of 20 independently and randomly chosen initial populations of 500 individuals each. The threshold G for each newborn differed from the value of its parent by a random number having a standard normal distribution. The population size was limited by its initial number of 500 via the stochastic elimination of sufficient newborns to ensure a constant population size of 500. The results, for each simulation, were the median values of the four life history traits, i.e. birthweight, adult weight, age at maturity and life expectancy, of 10 000 individuals obtained by the end of 25 000 iterations. Comparative analysis

Food, a

Parasitical stress, qp

Accidental mortality, qe

Individual, i

Birthweight, Bi Size, Wi (t) Age, Ai (t) Energy, Ei (t) Control , ui (t)

Reproduction, Bi (t)

Mortality, Qi (t)

International data on mean birthweight and percentage of preterm birth (i.e. below 37 weeks) were obtained from a database available from the World Health Organ5 ization (WHO) (2000). These data come from various sources (e.g. scientific journals, reports from ministries of health...) and combine male and female data. When several values were given for the same country (i.e. several cities), we used the mean of these values. In other situations, data refer to a limited part of a population (e.g. a given city), and thus an important assumption we made is that they are representative of the country. It is frequently argued that comparative analyses using information from different sources may be inappropriate because data have been collected with different methods, or they come from different sources. Although the argument is always applicable when no significant result is detected (i.e. data are not precise enough to detect a potentially significant result), it is unlikely to be relevant when significant trends are found, as a biological tendency has no a priori reason to be correlated to background noise in the data set (MÜller, 1997; Lawton, 1999). Because alimentary conditions are known to strongly affect birthweight, we also considered the mean calorie consumption per person per day. This information

J. EVO L. BIO L . d o i :10 . 1 111 /j.142 0-910 1.20 04.00 705 . x ª 20 04 B L A C K WELL P UB LIS H ING L TD

PR OO F


Human birthweight evolution

5

(a)
0.008 2.5 3.0

A C

0.007

0.006

0.005

0.004

0.003

0.002

0.001

0 0 0.002 0.004 0.006

0.008

Birthweight-dependent stress

(b)

UN CO RR EC TE D
4.25 4.00 3.75

4.50

Fig. 2 (a) The contour plot of modelled optimal birthweight against birthweightdependent and birthweight-independent components of environmental stress. The values of birthweight are shown by the numbers near contour lines and by the saturation of contour areas. The bold curves A, B, and C correspond to different scenarios of simultaneous increase of two components (see text). (b) The dependencies of birthweight on birthweight-dependent component of stress corresponding to three scenarios A, B, and C of joint increase of two components of stress shown in (a).

Birthweight

3.50

3.25

3.00

2.75

2.50

0

0.002

0.004

0.006

0.008

Birthweight-dependent stress

was obtained from food balance sheets from the World Atlasª (1992). Data on female stature were taken from ´ ´ Guegan et al. (2001). As in previous papers (Guegan et al., 2000, 2001; Thomas et al., 2000, 2001), disease occurrence in the different countries were compiled from two international health care databases, i.e. the Centres for Disease Control and Prevention (Atlanta, USA at http://www.cdc.gov/) and the World Health Organization (Geneva, Switzerland at http://www.who.int/). We collected data for a set of 16 categories of human parasitic and infectious diseases known to affect human survival (typhoid, hepatitis A, hepatitis B, malaria, schistosomiasis, filariosis, meningococcosis, yellow fever, dengue fever, cholera, trypanosomiasis, dracunculosis, chagas disease, Lyme disease, cutaneous leishmaniosis and visceral leishmaniosis). Based on this information, we calculated the disease load as the total number of diseases

for each country with a maximum disease load equal to 13. Because data on all variables were not always available, our study was based on different numbers of countries for different analyses ­ from 51 to 130 (the most complete list included 19 European countries, 37 countries from Asia, 40 from Africa and 37 from America).

Statistical analyses

To analyse the relationship between birthweight and parasitic/infectious diseases, we first performed polynomial regressions of crude values of birthweight on the number of infections to capture nonlinearity, and retained the regression which was the most significant. In a second step, we performed the same analyses using birthweight corrected for both the percentage of preterm children and for the mean number of calories per person

J. EVO L. BIO L . d oi:10 . 1 111 /j. 1 420 -910 1.200 4.007 05.x ª 20 04 BL ACK WELL P UBL I S H ING L TD

PR OO F
3.5 4.0 4.5 0.010 0.012 0.014 0.016 0.018

Birthweight-independent stress

B

0.020

B

C

A

0.010

0.012

0.014

0.016

0.018

0.020


6

F. THOM AS ET AL .

(a)

0.008

30

40

0.007

0.006

0.005

0.004 60 0.003

0.002

0.001

0 0 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.018 0.018

0.020

Birthweight-dependent stress

(b)
0.008

0.007

0.006

0.005

0.004

0.003

0.002

0.001

per day using residual values from multiple regressions. Because female stature and obstetrical performance are ´ usually linked (see Guegan et al., 2001), we also examined the relationship between mean female stature and parasitic/infectious diseases by performing polynomial regressions. To find confidence limits for the position of the minimum of the quadratic regression, we used a bootstrap method (Efron, 1979) by calculating estimates of this position for 1000 random re-samples from the original data.

Results

Evolutionary modelling

The essence of the results is presented in Fig. 2a in the form of the dependence of optimal birthweight, wb, on

UN CO RR EC TE D
0 0 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.018 0.018 0.020

17

19

21

23

Birthweight-independent stress

Birthweight-dependent stress

birthweight-dependent parasitic and birthweight-independent environmental components of mortality, qP and qe. This dependence is fairly well approximated by a quadratic polynomial function of the form: wb ¼ 3:56 × 71:90qp þ 166:16qe þ 743:95q2 p þ 1286:26qP qe × 877:20q2 e

The contour lines in Fig. 2a are contour lines of this polynomial. The general tendency consists of an increase in the optimal birthweight with increasing parasite stress and decreasing environmental stress. However, as qP and qe are positively correlated in reality (Thomas et al., 2000), it is important to consider the evolution of birthweight along curves on the plane (qp, qe) corresponding to simultaneous increases of qp and qe.

J. EVO L. BIO L . d o i :10 . 1 111 /j.142 0-910 1.20 04.00 705 . x ª 20 04 B L A C K WELL P UB LIS H ING L TD

PR OO F
Fig. 3 (a) The contour plot of modelled optimal adult weight against birthweightdependent and birthweight-independent components of environmental stress. The values of adult weight are shown by the numbers near contour lines and by the saturation of contour areas. (b) The contour plot of modelled optimal age at maturity against birthweight-dependent and birthweight-independent components of environmental stress. The values of age at maturity are shown by the numbers near contour lines and by the saturation of contour areas.

50

Birthweight-independent stress


Human birthweight evolution

7

Residuals of weight at birth on birth precocity and calorie s

We considered three scenarios of correlated increases in qp and qe, illustrated by the three bold curves in Fig. 2a. First, (A), the optimal birthweight monotonically declines if the rate of increase of the parasite stress component is inferior to the rate of increase of the environmental stress component. Then, (B), the optimal birthweight monotonically rises if the rate of increase of the parasite stress component is superior. Finally, (C), the optimal birthweight first declines and then rises if the rate of increase of the environmental stress component is first superior and then inferior compared with the rate of increase of the parasite stress component. The change of birthweight along these three curves is shown on Fig. 2b. We see indeed that in case (A) the optimal birthweight monotonically declines with increasing qp, in the second case, (B), it rises, and in the third case, (C), it first declines and then rises. The patterns of change of evolutionary optimal values of adult weight and age at maturity as dependent on qp and qe are shown on Fig. 3a and b, respectively. Comparative analysis

(a)
3.6 3.5

3.4

3.3

3.2 Weight at birth

3.1

3.0

2.9

2.8

2.7

2.6

2.5 2 3

(b)

0.5

0.4

0.3

From data collected for 130 countries, birthweight, wb, appears to depend nonmonotonically on the number of infections, i (Fig. 4a). This dependency is well fitted by a quadratic parabola function wb ¼ 3:744 þ 0:163i × 0:00885i
2

Both the coefficients of i and i2 are significant at the levels P < 0.000001 and P ¼ 0.000001, respectively, illustrating that the dependency between birthweight and infection is significantly nonlinear. The minimum for this parabola is attained at the number of infections equalling imin ¼ )0.5 · ()0.163/ 0.00885) % 9.2 that is less than the greatest numbers of infections, i.e. i ¼ 12, encountered in real data. We calculated the 95% bootstrap confidence interval for imin,[8.7 < imin < 10.1], which falls inside the range of observed number of infections, i.e. between 3 and 12. To exclude the possibility that the dependence of birthweight on infections is not quadratic but simply asymptotically decreasing, we estimated the significance of linear regression of birthweight on infections for the most parasitized countries (with more than nine infections). We obtained the following equation based on the data for 30 countries wb ¼ 2:404 × 0:056i which is statistically significant at the level P ¼ 0.037. We eliminated the negative effect of birth prematurity and the positive effect of food supply on birthweight (Matsuda et al., 1995; Albertsson-Wikland et al., 1998) by repeating the same analysis for the residuals, rb, of birthweight from the linear regression between the percentage of children incubated for <37 weeks and the mean number of calories per person per day. The

UN CO RR EC TE D
0.2 0.1 0.0 ­0.1 ­0.2 ­0.3 ­0.4 ­0.5 2 3 4

(c)

170.0

167.5

165.0

162.5

160.0

Female adult size

157.5

155.0

152.5

150.0

147.5

145.0

2

3

Fig. 4 (a) The observed dependency of birthweight on number of infections. (b) The observed dependency of the residuals of birthweight from its linear regression on the per cent of children incubated <37 weeks and the number of calories per person per day on the number of infections. (c) The observed dependency of adult female stature on number of infections.

J. EVO L. BIO L . d oi:10 . 1 111 /j. 1 420 -910 1.200 4.007 05.x ª 20 04 BL ACK WELL P UBL I S H ING L TD

PR OO F
4 5 6 7 8 9 10 11 12 13 Number of infections
5 6 7 8 9 10 11 12 13 Number of infections

4

5

6

7

8

9

10

11

12

13

Number of infections


8

F. THOM AS ET AL .

rb ¼ 0:286 þ 0:0878i × 0:00554i

2

with the coefficients of i and i2 being significant at the levels P < 0.0018 and P ¼ 0.0037 respectively and with imin % 7.9. The nonmonotonic nature of the dependence on the number of infections is again well expressed (see Fig. 4b) and the bounds of the 95% confidence limit for imin,[7.2 < imin < 9.0], again do not exceed the range of observed numbers of infections. Secondly, taking into account that the birthweight is significantly correlated with adult female stature (R ¼ 0.50, P ¼ 0.00068), we analysed the dependency of adult female stature on the number of infections. As a result, we obtained for adult female stature a nonmonotonous dependency on the number of infections based on the data for 51 countries wa ¼ 173:63 þ 4:604i × 0:272i2 with imin % 8.5 (see Fig. 4c). The coefficients of i and i2 in this equation are significant at P ¼ 0.00079 and 0.0025 respectively, and the bounds of the 95% confidence limit for imin, [8.0 < imin < 9.2], do not exceed the range of observed numbers of infections, which confirms indirectly the existence of a minimum in the dependence of birthweight on infection.

Discussion

Investigating the evolution of behavioural and/or life history traits in humans is often limited by the fact that one must assume that current environmental conditions provide, at least partially, a reliable picture of the selective landscape experienced by human populations in the past (Barret et al., 2002). Numerous countries have recently (some 300 years ago in some developed countries and in less developed countries about 1920) undergone what are known as the demographic- and the epidemiologic-transitions (see Coale & Treadway, 1986). These transitions correspond to major changes whereby both fertility and mortality (including infectious and parasitic disease mortality) decreased whereas at the same time, we have seen an increase in the relative prevalence of degenerative diseases in countries where mortality has declined. A major assumption made in this study is that both current human life history traits and major environmental characteristics (i.e. the data we used) are reliable for testing our evolutionary predictions on birthweight. Although parasites are often harmful to their hosts, relatively few studies have illustrated how selection has favoured parental responses improving offspring resistance to infections, beyond selection for responses which alleviate the direct impact of parasites on infected hosts

UN CO RR EC TE D

J. EVO L. BIO L . d o i :10 . 1 111 /j.142 0-910 1.20 04.00 705 . x ª 20 04 B L A C K WELL P UB LIS H ING L TD

PR OO F

dependency of these residuals on the number of infections can be fitted by a quadratic regression based on the data for 89 countries

(Brambell, 1970; Sorci et al., 1994; Carlier & Truyens, 1995; Buechler et al., 2002). Parasitic and infectious diseases have been and continue to be a major cause of human mortality in the world, most particularly in tropical areas where many different agents are dreadful killers (Jackson, 2000). Because children with low birthweight are usually more susceptible to infections than others (see references in Introduction), we here predicted that natural selection should have operated on mechanisms (e.g. physiology, psychology, behaviour...) that enable human females to produce larger children when infection risks are extreme. Our study apparently supports this hypothesis since, once a threshold in the infection risk is reached (nine diseases in our study), birthweights significantly increase with the number of diseases present. Further studies are necessary to determine whether the simple accumulation of harmful diseases is, per se, responsible for this pattern, or whether it results from the particular effects of some diseases among those that appear in communities beyond the eight to nine diseases point. The two most important variables known to influence birthweight in the modern literature (i.e. gestation length and nutritive conditions) do not confound this result as the same pattern is observed after controlling statistically for these effects. However, other potential confounding variables do remain. For instance, available birthweight data records only live births, hence differential mortality of smaller foetuses in parasiticallydiverse countries could bias birthweight towards larger babies. Further information would be necessary to clarify the exact links between birthweight and resistance/ susceptibility to parasitic and infectious diseases. Indeed, many components of the immune system develop early in foetal life (Good & Gabrielsen, 1964) and maternal malnutrition is generally observed to have greater effect on thymic and lymphoid tissue growth than on other organs (Owens et al., 1989; Moore et al., 1999). In this context, a higher birthweight per se may not directly confer a higher capacity to resist infections, instead it would simply correlate with nutritive conditions met by the foetus when developing his/her immune organs. In a previous study considering 150 different countries, ´ Guegan et al. (2001) showed that human fertility increases with the diversity of parasitic and infectious diseases. The present work suggests that improving the chance of offspring survival is also likely to be a crucial component of individual fitness when parasitic pressures increase. Because fertility in heavily parasitized countries is usually very high for most women, fitness variation between mothers are more likely to result from variation in the survival of the offspring produced than on their number (Strassmann & Gillespie, 2002). In summary, ´ these results and those of Guegan et al. (2001) suggest that in human populations living under severe parasitic constraints, natural selection has favoured increased reproductive investment, represented by increases in


Human birthweight evolution

9

J. EVO L. BIO L . d oi:10 . 1 111 /j. 1 420 -910 1.200 4.007 05.x ª 20 04 BL ACK WELL P UBL I S H ING L TD

UN CO RR EC TE D

both the size and number of offspring (all things being equal). An increase in reproductive investment in response to raised (parasite-driven) mortality is in keeping with basic tenets of life history theory (Stearns, 1992). Furthermore our data points to a partition of increased investment into both number and size of offspring. Here, we argue that the increase in size is a direct selective response to the dependency of parasitelinked fitness losses on birthweight. However our results cannot exclude an alternative hypothesis that the global increase in reproductive budget is invested in multiple dimensions of reproductive output (size, number and potentially other dimensions such as lactation) to counter diminishing returns in any one dimension of investment (Stearns, 1992; Mangel et al., 1994). Further theoretical studies would be necessary to understand the trade-offs retained by selection between offspring number and offspring size for different scenarios of infection exposure (e.g. kinds of parasitic and infectious diseases, synergic effects among parasite groups, virulence, periodicity of the exposures). Interestingly female stature is positively correlated to birthweight, displaying the same pattern with regard to parasitic and infectious diseases, in particular it also increases after the threshold of eight to nine diseases is reached. These correlations are observed both in the data (compare Fig. 4a with Fig. 4c) and in the model (compare Fig. 2a with Fig. 3a). The simplest explanation for this correlation is that conditions that influence foetal growth also influence subsequent growth later in life. However, as we can see in Fig. 3b, increasing parasitic stress is accompanied with increasing age of maturity that allows to bear children of greater birthweight in severe conditions (see also Restif et al., 2001). In fact, there may be different explanations for this correlation between female stature and birthweight. We assumed in the model that increasing birthweight decreases the risk of death from infections. In our model, the selection for an increased adult size is rather secondary: larger adult sizes are necessary for obtaining higher birthweights. Alternatively, we could assume that higher adult sizes decrease the risk of death from infections. We would expect in this case a selection for a larger adult size in the most infected areas. This could be achieved by producing children with higher birthweights. Although we prefer the first hypothesis, both are theoretically possible. Also other phenomena are likely to occur. Indeed, knowing that there is a strong correlation between female height and risks of difficult childbirth (Camilleri, 6 1981; Parsons et al., 1989; Tsu, 1992; Van Roosmalem & Brand, 1992; Kwawukume et al., 1993; Moller & Lindmark, 1997), this correlation may result from adaptive adjustments of female morphology to the size of the children they produce. Thus, in heavily parasitized countries, not only high fecundities would exert selective ´ pressures for large stature in women (see Guegan et al.,

2001), but the need for producing larger children would also influence positively the evolution of this trait. Alternatively, we cannot exclude that selection in parasitized countries favours (for reasons that remain to be determined) adult individuals with large statures. Because of the positive relationship between the stature of females and the birthweight of their children (see references in Introduction), higher birthweights in parasitized areas could be a consequence of the former selection without the need to invoke an advantage of large babies over smaller one. Further studies are thus necessary to determine which trait(s) are the target(s) of the selection. Irrespectively of what is the main target of selection (birthweight or adult size), our modelling predicts that the observed tendency for the increase of both these parameters in the countries with highest infections can occur only if in the most dangerous conditions the importance of accidents as compared with infections becomes relatively less important. Additional data and their analysis are necessary to confirm or refute this prediction, as well as information on the role played by the age at first reproduction in these processes. Another important result of this study is that the largest birthweights are observed in countries with no or few harmful parasites. We predicted this result because of two interacting phenomena. First, knowing that these countries usually correspond to industrialized countries, it can be first argued that large amounts of nutritive resources are available to pregnant women. A second reason is that natural selection should have favoured individuals producing larger children because of their reduced probability of developing a chronic disease later in life compared with individuals who had a lower birthweight (see references in Introduction). Knowing that in rich countries, most deaths occur as a result of somatic diseases, such as cancer, strokes and heart disease (40% of all deaths in developing countries vs. 75% in industrialized countries, WHO, 1996 estimates), these causes of mortality/morbidity appear likely to be a major source of fitness variation between individuals. Thus, as both selective and environmental pressures lead in the same direction, it is unsurprising that birthweights reach the highest values in the most economically developed countries. Further investigations are however needed to valid our conclusions since, as far as we know, the mechanism of the relationship of birthweight to later susceptibility to chronic disease is unclear and still debated. If for instance mutations cause both the low birthweight and the later susceptibility to chronic diseases, maternal manipulation of birthweight would have no effect. Interestingly, once we correct birthweight for the influence of nutrition and of prematurity, mean birthweights in low- and in highly-infected countries become on average similar, confirming the preponderant role of environmental (nongenetic) variables in generating birthweight variation between industrialized and developing countries. However, other parameters undoubtedly

PR OO F


10

F. THO M A S ET A L .

Acknowledgments
We thank M. Raymond, H. Harpending and one anonymous referee for comments on an earlier version of the manuscript. F. Thomas is supported by an `ACI jeunes Chercheurs'.

References

Abell, T.D., Baker, L.C. & Ramsey, Ch.N. 1991. The effects of maternal smoking on infant birth weight. Fam. Med. 23: 103­107. Albertsson-Wikland, K., Boguszewski, M. & Karlberg, J. 1998. Children born small-for-gestational age: postnatal growth and hormonal status. Hormone Res. 49( Suppl. 2): 7­13. Barker, D., Osmond, C., Golding, J., Kuh, D. & Wadsworth, M.E.J. 1989. Growth in utero, blood pressure in childhood and adult life, and mortality from cardiovascular disease. BMJ 298: 564­567. Barret, L., Dunbar, R. & Lycett, J. 2002. Human Evolutionary Psychology. Palgrave, London, UK. Bereczkei, T. 1993. r-Selected reproductive strategies among Hungarian gypsies: a preliminary analysis. Ethol. Sociobiol. 14: 71­88. Bereczkei, T., Hofer, A. & Ivan, Z. 2000. Low birth weight, maternal birth-spacing decisions and future reproduction. Hum. Nat. 11: 183­205. Blurton Jones, N. 1978. Natural selection and birthweight. Ann. Hum. Biol. 5: 487­489. Bogin, B. & Smith, B.H. 2000. Evolution of the human life cycle. In: Human Biology, an Evolutionary and Biocultural Perspective (S. /forenames> Stinson, B. Bogin, R. Huss-Ashmore & D. O'Rourke, eds), pp. 377­424. Wiley-Liss, USA Brabin, B. 1991. An assessment of low birthweight risk in primiparae as an indicator of malaria control in pregnancy. Int. J. Epidemiol. 20: 276­283. Brambell, F.W.R. 1970. The Transmission of Passive Immunity from Mother to Young. Elsevier, New York. Buechler, K., Fitze, P.S., Gottstein, B., Jacot, A. & Richner, H. 2002. Parasite-induced maternal response in a natural bird population. J. Anim. Ecol. 71: 247­252. Bukenya, G., Barnes, T. & Nwokolo, N. 1991. Low birthweight and acute childhood diarrhoea: evidence of their association in an urban settlement of Papua New Guinea. Ann. Trop. Paed. 11: 357­362.

UN CO RR EC TE D

J. EVO L. BIO L . d o i :10 . 1 111 /j.142 0-910 1.20 04.00 705 . x ª 20 04 B L A C K WELL P UB LIS H ING L TD

PR OO F

intervene since significant birthweight variations (total data set) remain after controlling for the previous effects. In accordance with our predictions, peak values are observed for low- and for highly-parasitized countries, respectively. In conclusion, and to summarize, although a large number of nongenetic variables certainly contribute to explaining variability in birthweight worldwide, our study suggests that a significant part of the variability on this trait results from adaptive responses to local selective pressures. In such a context, parasitic pressures apparently played a significant role. Further analyses that include other kinds of pathogen (e.g. viruses) are necessary to confirm our findings.

Camilleri, A.P. 1981. The obstetric significance of short stature. Eur. J. Obstet. Gynecol. Reprod. Biol. 12: 347­356. Carlier, Y. & Truyens, C. 1995. Influence of maternal infection on offspring resistance towards parasites. Parasitol. Today 11: 94­99. Cerqueiro, M., Murtagh, P., Halac, A., Avila, M. & Weissenbacher, M. 1990. Epidemiologic risk factors for children with acute lower respiratory tract infections in Buenos Aires, Argentina: a matched case­control study. Rev. Infect. Dis. 12: 1021­1028. Chandra, R.K. 1997. Nutrition and the immune system: an introduction. Am. J. Clin. Nutr. 66: 460­463. Chandra, R.K. 1999. Nutrition and immunology: from the clinic to cellular biology and back again. Proc. Nutr. Soc. 58: 681­683. Cicognani, A., Alessandroni, R., Pasini, A., Pirazzoli, P., Cassio, A., Barbieri, E. & Cacciari, E. 2002. Low birth weight for gestational age and subsequent male gonadal function. J. Pediat. 141: 376­380. Coale, A.J. & Treadway, R. 1986. A summary of the changing distribution of overall fertility, marital fertility and the proportion married in the provinces of Europe. In: The Decline of Fertility in Europe (A. J. Coale & S. C. Watkins, eds), pp. 31­ 181. Princeton University Press, Princeton. Cramer, J.C. 1995. Racial and ethnic differences in birthweight: the role of income and financial assistance. Demography 32: 231­248. Defo, B.K. & Partin, M. 1993. Determinants of low birthweight: a comparative study. J. Biosoc. Sci. 25: 87­100. Efron B. 1979. Bootstrap methods: another look at the jacknife. Ann. Stat. 7: 1­26. Emanuel, I., Filakti, H., Alberman, E. & Evans, S.J.W. 1992. Intergenerational studies of human birthweight from the 1958 birth cohort. 1. Evidence for a multigenerational effect. Brit. J. Obst. Gynecol. 99: 67­74. Falkner, B. 2002. Birth weight as a predictor of future hypertension. Am. J. Hypert. 15: 43­45. Ferguson, A.C. 1978. Prolonged impairment of cellular immunity in children with intrauterine growth retardation. J. Pediatr. 93: 52­56. Fonseca, W., Kirkwood, B.R., Victora, C.G., Fuchs, S.R., Flores, J.A. & Misago, C. 1996. Risks factors for childhood pneumonia among the urban poor in Fortaleza, Brazil: a case­control study. Bull. World Health Organ. 74: 199­208. Futuyma, D. 1998. Evolutionary Biology, 3rd edn. Sinauer, 7 Sunderland, MA, USA. Gavrilov, L.A. & Gavrilova, N.S. 1991. The Biology of Life Span: A Quantitative Approach. Harwood Academic Publishers, New York. Godfrey, K.M. & Barker, D.J. 2000. Fetal nutrition and adult disease. Am. J. Clin. Nutr. 71: 1344­1352. Gompertz, B. 1825. On the nature of the function expressive of the law of human mortality and on a new mode of determining life contigencies. Phil. Trans. Roy. Soc., A, 115: 513­585. Good, R.A. & Gabrielsen, A.E. 1964. The Thymus in Immunobiology. Structure, Function and Role in Disease. Harper and Row, London. ´ Guegan, J.F., Teriokhin, A. & Thomas, F. 2000. Human fertility variation, size-related obstetrical performance and the evolution of sexual stature dimorphism. Proc. Roy. Soc. Lond. B. 267: 2529­2536.


Human birthweight evolution

11

J. EVO L. BIO L . d oi:10 . 1 111 /j. 1 420 -910 1.200 4.007 05.x ª 20 04 BL ACK WELL P UBL I S H ING L TD

UN CO RR EC TE D

´ Guegan, J.F., Thomas, F., de Meeus, T., Hochberg, M.E. & ¨ Renaud, F. 2001. Disease diversity and human fertility. Evolution 55: 1308­1314. Heeb, P., Werner, I., Kolliker, M. & Richner, H. 1998. Benefits of ¨ induced host responses against an ectoparasite. Proc. Roy. Soc. 8 Lond. B 265: 51­56. Ittiravivongs, A., Songchitratna, K., Ratthapalo, S. & PattaraArechachai, J. 1991. Effect of low birthweight on severe childhood diarrhea. South. As. J. Trop. Med. Pub. health 22: 557­562. Jackson, F.L.C. 2000. Human adaptations to infectious disease. In: Human Biology, an Evolutionary and Biocultural Perspective (S. Stinson, B. Bogin, R. Huss-Ashmore & D. O'Rourke, eds), pp. 273­293. Wiley-Liss, USA. Klebanoff, M.A., Mednick, B.R., Schulsinger, C., Secher, N.J. & Shiono, P.H. 1998. Father's effect on infant birthweight. Am. J. Obstet. Gynecol. 178: 1022­1026. Kleinman, J. & Kessel, S. 1987. Racial differences in low birth weight: trends and risk factors. New Engl. J. Med. 317: 749­753. Kopp, C.B. 1990. Risks in infancy: appraising the research. Merril-Palmer Quaterly 36: 117­140. Koupilova, I., Rahu, K., Rahu, M., Karro, H. & Leon, D.A. 2000. Social determinants of birthweight and length of gestation in Estonia during the transition to democracy. Int. J. Epidemiol. 28: 1088­1095. Kozlowski, J. & Teriokhin, A.T. 1999. Energy allocation between growth and reproduction: pontryagin maximum principle solution for the case of age- and season dependent mortality. Evol. Ecology Res. 1: 423­441. Kwawukume, E.Y., Ghosh, T.S. & Wilson, J.B. 1993. Maternal height as a predictor of vaginal delivery. Int. J. Gynaecol. Obstet. 41: 27­30. Langhoff-Roos J., Lindmark, G., Gustavsson, K.H., Gebremedhin, M. & Meirik, O. 1987. Relative effect of parental birthweight on infant birthweight at term. Clin. Genet. 32: 240­ 248. Lawton, J.H. 1999. Are there general laws in ecology? Oikos 84: 177­192. Le, L.T., Kiely, J.L. & Schoendorf, K.C. 1996. Birthweight outcomes among Asian American and Pacific subgroups in the United states. Int. J. Epidemiol. 25: 973­979. Lummaa, V. & Clutton-Brock, T. 2002. Early development, survival and reproduction in humans. TREE 17: 141­147. Makeham, W.M. 1860. On the low of mortality and the construction of annuity tables. J. Inst. Actuaries 8: 301­ 310. Mangel, M. 1990. Evolutionary optimization and neural network models of behavior. J. Math. Biol. 28: 237­256. Mangel M. & Clark, C. 1988. Dynamical Modeling in Behavioral Ecology. Princeton Univ. Press, Princeton. Mangel M., Rosenheim J.A. & Adler, F.R. 1994. Clutch size, offspring performance, and intergenerational fitness. Behav. Ecol. 5: 412­417. Matsuda, S., Hiroshige, Y., Furuta, M., Doi, T., Sone, T. & Kahyo, H. 1995. Geographic differences in seasonal variation of mean birth weight in Japan. Hum. Biol. 67: 641­ 656. McNamara, J.M. & Houston, A. 1996. I. State-dependent life histories. Nature 380: 215­221. MÜller, A.P. 1997. Development stability and fitness: a review. Am. Nat. 149: 916­932.

Moller, B. & Lindmark, G. 1997. Short stature: an obstetric risk factor? A comparison of two villages in Tanzania. Acta Obstet. Gynecol. Scand. 76: 394­397. Moore, S.E., Cole, T.J., Collinson, A.C., Poskitt, E.M.E., McGregor, I.A. & Prentice, A.M. 1999. Prenatal or early postnatal events predict infectious deaths in young adulthood in rural Africa. Int. J. Epidemiol. 28: 1088­1095. Mylius, S.D. & Diekmann, O. 1995. On evolutionary stable life histories, optimization and the need to be specific about density dependence. Oikos 74: 218­224. Nafstad, P., Samuelsen, S.O., Irgens, L.M. & Bjerkedal, T. 2002. Birth weight and hearing impairment in Norwegians born from 1967 to 1993. Pediatrics 110: 16­21. Owens, J.A., Owens, P.C. & Robinson, J.S. 1989. Experimental fetal growth retardation: metabolic and endocrine aspects. In: Advances in Fetal Physiology (P. D. Gluckman, B. M. Johnston & P. W. Nathanielsz, eds), pp. 263­286. Perinatology Press, Ithaca, New York. Parsons, M.T., Winegar, A., Siefert, L. & Spellacy, W.N. 1989. Pregnancy outcomes in short women. J. Reprod. Med. 34: 357­361. Perrin, N. & Sibly, R.M. 1993. Dynamic models of energy allocation and investment. Ann. Rev. Ecol. Syst. 24: 379­410. Restif, O., Hochberg, M.E. & Koella, J.C. 2001. Virulence and age at reproduction: new insights into host­parasite coevolution. J. Evol. Biol. 14: 967­979. Robertson, C.M.T., Etches, P.C. & Kyle, J.M. 1990. Eight-year school performance and growth of preterm, small for gestional age infants. J. Pediat. 116: 16­19. Roff, D.A. 1992. The Evolution of Life Histories: Theory and Analysis. Chapman and Hall, New York. Shiono, P.H., Rauh, V.A., Park, M., Lederman, S.A. & Zuskar, D. 1997. Ethnic differences in birthweight: the role of lifestyle and other factors. Am. J. Pub. Health 87: 787­ 794. Sorci, G., Massot, M. & Clobert, J. 1994. Maternal parasite load increases sprint speed and philopatry in female offspring of the common lizard. Am. Nat. 144: 153­164. Stearns, S. 1992. The Evolution of Life-histories; Theory and Analysis. Oxford University Press, Oxford. Stearns, S.C. & Hoekstra, R.F. 2002. Evolution, an Introduction. Oxford University Press, Oxford, UK. Stinson, S. 2000. Growth variation: biological and cultural factors. In: Human Biology, an Evolutionary and Biocultural Perspective (S. Stinson, B. Bogin, R. Huss-Ashmore & D. O'Rourke, eds), pp. 425­463. Wiley-Liss, USA. Strassmann, B.I. & Gillespie, B. 2002. Life history theory, fertility and reproductive success in humans. Proc. R. Soc. Lond. B. 269: 553­562. Teriokhin, A.T. 1998. Evolutionarily optimal age schedule of repair: computer modeling of energy allocation between current and future survival and reproduction. Evol. Ecol. 12: 291­307. Teriokhin, A.T. 2002. Models of Competition: Population Dynamics and Phenotype Evolution (in Russian). MaxiPress, Moscow. ´ Thomas, F., Teriokhin, A., Renaud, F., De Meeus, T. & Guegan, ¨ J.F. 2000. Human longevity at the cost of reproductive success: evidence from global data. J. Evol. Biol. 13: 409­414. ´´ ´ Thomas, F., Renaud, F., Benefice, E., de Meeus, T. & Guegan, ¨ J.F. 2001. International variability on ages at menarche and menopause: patterns and main determinants. Hum. Biol. 73: 271­290.

PR OO F


12

F. THO M A S ET A L .

Towne, B., Guo, S., Roche, A.F. & Siervogel, R.M. 1993. Genetic analysis of patterns of growth in infant recumbent length. Hum. Biol. 65: 977­989. Trivers, R.L. 1974. Parent­offspring conflict. Am. Zool. 14: 247­262. Tsu, V.D. 1992. Maternal height and age: risk factors for cephalopelvic disproportion in Zimbabwe. Int. J. Epidemiol. 21: 941­946. Tulassay, T. & Vasarhelyi, B. 2002. Birth weight and renal function. Curr. Opin. Nephr. Hypert. 11: 347­352. Van Roosmalem, J. & Brand, R. 1992. Maternal height and the outcome of labor in rural Tanzania. Int. J. Gynaecol. Obstet. 37: 169­177. Vangen, S., Stoltenberg, C., Skjaerven, R., Magnus, P., Harris, J.R. & Stray-Pedersen, B. 2002. The heavier the better? Birthweight and perinatal mortality in different ethnic groups. Int. J. Epidemiol. 31: 654­660. Victora, C.G., Smith, P.G., Vaughan, J.P., Nobre, L.C., Lombardi, C., Teixeira, A.M., Fuchs, S.M., Moreira, L.B., Gigante, L.P. & Barros, F.C. 1988. Influence of birth weight on mortality from infectious diseases: a case­control study. Pediatrics 81: 807­811.

Received 16 September 2003; revised 18 December 2003; accepted 19 December 2003

UN CO RR EC TE D
J. EVO L. BIO L . d o i :10 . 1 111 /j.142 0-910 1.20 04.00 705 . x ª 20 04 B L A C K WELL P UB LIS H ING L TD

PR OO F

Victora, C.G., Smith, P.G., Vaughan, J.P., Nobre, L.C., Lombardi, C., Teixeira, A.M., Fuchs, S.M., Moreira, L.B., Gigante, L.P. & Barros, F.C. 1989. Infants feeding and deaths due to diarrhea. Am. J. Epidemiol. 129: 1032­1041. Victora, C.G., Barros, F.C., Kirkwood, B.R. & Vaughan, J.P. 1990. Pneumonia, diarrhea, and growth in the first 4y of life: a longitudinal study of 5914 urban Brazilian children. Am. J. Clin. Nutr. 52: 391­396. Wells, J.C.K. 2002. Thermal environment and human birth weight. J. Theor. Biol. 214: 413­425. West, G.B., Brown, J.H. & Enquist, B.J. 1999. The fourth dimension of life: fractal geometry and allometric scaling of organisms. Science 284: 1677­1679. WHO 1996. Noncommunicable Disease, WHO Experts Warn Against Inadequate Prevention, Particularly in Developing Countries. Fact Sheet N 106. Zhang, J. & Savitz, D.A. 1992. Preterm birth subtypes among blacks and whites. Epidemiology 3: 428­433.


Author Query Form
Journal: Article: JEB 705

Dear Author, During the copy-editing of your paper, the following queries arose. Please respond to these by marking up your proofs with the necessary changes/additions. Please write your answers on the query sheet if there is insufficient space on the page proofs. Please write clearly and follow the conventions shown on the attached corrections sheet. If returning the proof by fax do not write too close to the paper's edge. Please remember that illegible mark-ups may delay publication. Many thanks for your assistance.

Query reference Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8

Query Au: Langhoff-Roos, 1987 has been changed to Langhoff-Roos et al., 1987 so that this citation matches the list Au: Emanuel, 1992 has not been included in the list, please supply publication details. Au: Klebanoff, 1998 has been changed to Klebanoff et al., 1998 so that this citation matches the list Au: McNamara & Huston, 1996 has been changed to McNamara & Houston, 1996 so that this citation matches the list Au: World Health Organization (2000) has not been included in the list, please supply publication details. Au: Parsons, 1989 has been changed to Parsons et al., 1989 so that this citation matches the list Au: Year `1998' inserted from text citation. Is this ok? ¨ Au: Heeb, Werner, Kolliker, Richner (1998) not cited. Please cite reference in text or delete from the list.

Remarks


ïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïï

MARKED PROOF

ïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïïï

Please correct and return this set

Please use the proof correction marks shown below for all alterations and corrections. If you wish to return your proof by fax you should ensure that all amendments are written clearly in dark ink and are made well within the page margins.
Instruction to printer Leave unchanged Insert in text the matter indicated in the margin Delete Delete and close up Substitute character or substitute part of one or more word(s) Change to italics Change to capitals Change to small capitals Change to bold type Change to bold italic Change to lower case Change italic to upright type Insert `superior' character Textual mark under matter to remain through matter to be deleted through matter to be deleted through letter or through word Marginal mark Stet New matter followed by

New letter or new word

under matter to be changed under matter to be changed under matter to be changed under matter to be changed under matter to be changed Encircle matter to be changed (As above) through character or where required Insert `inferior' character (As above) Insert full stop (As above) Insert comma (As above) Insert single quotation marks (As above) Insert double quotation (As above) marks Insert hyphen (As above) Start new paragraph No new paragraph Transpose Close up linking letters Insert space between letters between letters affected Insert space between words between words affected Reduce space between letters between letters affected Reduce space between words between words affected

under character e.g. over character e.g. and/or and/or