The effect of ethnic admixture and parental lineage on birth weight at high altitude

Material Information

The effect of ethnic admixture and parental lineage on birth weight at high altitude
Bennett, Adam Forrest
Publication Date:
Physical Description:
ix, 67 leaves : ; 28 cm


Subjects / Keywords:
Birth weight -- Bolivia -- La Paz ( lcsh )
Ethnology -- Bolivia -- La Paz ( lcsh )
Altitude, Influence of -- Bolivia -- La Paz ( lcsh )
Acclimatization -- Bolivia -- La Paz ( lcsh )
Acclimatization ( fast )
Altitude, Influence of ( fast )
Birth weight ( fast )
Ethnology ( fast )
Bolivia -- La Paz ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 58-67).
General Note:
Department of Anthropology
Statement of Responsibility:
by Adam Forrest Bennett.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
63787564 ( OCLC )
LD1193.L43 2004m B46 ( lcc )

Full Text
Adam Forrest Bennett
B.A., Northwestern University, 1996
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Arts

This thesis for the Master of Arts
degree by
Adam Forrest Bennett
has been approved
David Tracer


Bennett, Adam Forrest (M.A., Anthropology)
The Effects of Ethnic Admixture and Parentage on Birth Weight at High Altitude
Thesis directed by Professor Loma G. Moore
This study utilizes medical records from La Paz, Bolivia (3600m) to determine the
effect of ethnic admixture and parentage on birth weight at high altitude. Birth
records from 1385 singleton deliveries were classified by ethnicity. Birth weights
were compared among ethnicity groups using ANCOVA, categorical regression
techniques, and linear contrasts with SPSS. A birth weight gradient existed among
paternal and maternal groupings such that birth weight increased progressively with
the degree of Andean ancestry. Qualitatively, the lowest birth weights occurred in
babies whose mothers were Foreign, whereas the highest birth weights occurred in
babies whose fathers were Andean. Paternal and maternal ethnicity were highly
correlated and there were very few Foreign-Andean matings, illustrating the
occurrence of assortative mating in this population. Our findings show that birth
weight at high altitude is influenced by ethnicity and further, that this effect is dose-
dependent. In addition, the effect of maternal ancestry is greater than that of paternal
ancestry. Collectively, these results suggest that a genetic factor operates to protect
Andean populations from altitude-associated reductions in birth weight, enabling
adaptation to high altitude. They are also consistent with David Haigs genetic
conflict theory of pregnancy which posits that, while from the fathers perspective,
evolution favors larger offspring, the mother must balance present costs against
future reproduction and thereby reduces her somatic investment in any one fetus.
Given population differences in susceptibility to birth weight reduction at high
altitude, admixture among ethnic groups provides a unique means for identifying the
genetic and cultural factors involved. Future studies such as this one promise to be
informative for achieving a population-level approach to intergenerational conflicts
over resource allocation.

This abstract accurately represents the content of the candidates thesis. I
recommend its publication.

I wish to thank Loma G. Moore for her patient support of my research. I have
learned a great deal and gained confidence through working with her. I would also
like to acknowledge David Tracer and John Brett for their help in keeping my ship
righted. For their advice, consultation, and friendship I would like to thank Megan
Wilson, Wendy MacCannell, and Erica Ferro. For maintaining my sense of humor
as I waded through sometimes murky academic waters, Craig Kapral and Steve
Koesters friendship have been invaluable. Connie Turner deserves special regard,
because she has endured my presence and allowed me wonderful perspective.

1. INTRODUCTION...............................................1
2. BACKGROUND................................................6
Population factors based upon ethnicity and genetics...6
The high altitude environment..........................8
The social environment................................12
Life history theory...................................14
Parent-offspring conflict.............................19
Genomic imprinting....................................21
Evolutionary explanations.............................24
Genomic imprinting and placental physiology...........28
3. METHODS..................................................32
Data collections and definition.......................32
Analysis............................................ 36
4. RESULTS..................................................38
Distribution of matings...............................38

Differences in birth weight between ethnic groups....40
Gestational age......................................44
Maternal weight gain.................................47
Socioeconomic status.................................47
5. DISCUSSION...................................... 50

2.1 Birth weight by altitude for different ethnic groups.......................11
2.2 Paternal and maternal conflict over growth factors.........................23
2.3 Maternal placental circulation............................................31
3.1 Hispanic assignment maternal and paternal surnames........................34
4.1 Distribution of matings by maternal and paternal ethnicity................39
4.2 Birth weight classified by parental ethnic admixture.......................41
4.3 Birth weight classified by ethnicity in the mestizo group..................43
4.4 Relationship between parity and birth weight..............................46

2.1 Known imprinted genes in humans.........................................24
3.1 Variables assessed......................................................34
4.1 Mean birth weight values classified by parental ethnic category........40
4.2 Mean values for covariates, categorized by maternal ethnicity..........44
4.4 Maternal weight gain and socioeconomic status..........................48

Birth weight is an important predictor of both perinatal and adult morbidity and
mortality. Globally, low birth weight (below 2500 grams) has long been recognized
as one of the most significant contributors to neonatal and infant mortality (Keyes et
al 2003). More recently, low birth weight has also been linked to childhood and
adult onset diseases such as hypertension, diabetes, and coronary heart disease
(Barker 1999). Environmental factors, both physical and social, often affect birth
weight outcomes. Specifically, the high altitude environment presents a serious
challenge to fetal growth because of the lower oxygen availability. As a result of the
effects of this altitude-induced hypoxia upon fetal growth, infant mortality rates rise
with increasing altitude. Birth weights at high altitude are generally lower than those
at sea level, with the birth weight reduction at high altitude averaging 100 grams for
every 1000 meter gain (Jensen and Moore 1997).
However, specific population factors affect the degree to which high altitude
impacts birth weight. Population differences have been established as influencing
birth weight at high altitude through comparative research throughout the high
altitude regions of the world (Moore 2001). The foundational work for this thesis,
conducted by Loma G. Moore, sought to assess the influence of generational length

of high-altitude residence on altitude-associated reductions in infant birth weight.
Her results show that populations that have resided for a longer generational period
at high altitude demonstrate less altitude associated reductions in birth weight,
thereby potentially implicating an accumulated protective genetic effect (natural
selection in action) (Moore 2001).
If these population distinctions are taken as a proxy for genetic differences, we can
hypothesize that in admixted populations, the genetic influence on altitude-
associated reductions in birth weight would vary based upon an individuals dose
of genetic admixture. Ethnicity, as established through parental surnames, can be
considered a good measure of genetic admixture. The use of ethnicity to separate
populations is supported by Chakrabortys work in Bolivia, which has shown an
89% correlation between Aymara surnames and Amerindian genetic markers .
(Chakraborty 1989).
Utilizing the data from the medical records of 1385 women in La Paz, this thesis
will address how population differences in birth weight are affected by ethnic
admixture and parental lineage at high altitude. The analyses will seek to ascertain
what the effects of ethnic admixture by parental lineage are, whether they differ for
mothers and fathers, what might explain the differences, and how a population
perspective on ethnic admixture might be useful in understanding those differences.
In so doing, this work adds to the analyses developed by Moore and others in two
regards: by providing a more fine-scale assessment of the relationship between
ethnic admixture and birth weight at high altitude, and by comparing the relative
effects of admixture in the maternal vs. paternal lineages on birth weight.
Specifically, the analysis presented here will show that the effect of maternal ethnic
admixture on birth weight is greater than the paternal effect. Further, it will evaluate
the effectiveness of a population perspective for understanding the role of genetic

factors in influencing fetal growth, specifically that of genomic imprinting.
Genomic imprinting occurs when the expression of an allele is influenced by the
parent from which it has been received.
Methodologically, a population-level perspective allows for the assessment of broad
trends and potentially, the interaction between biological and cultural factors
affecting reproductive outcomes. Accordingly, we can view the biological and
cultural environments, and their confluence, as part of a causal web influencing
birth weight. More specifically, the hypoxic conditions of the high-altitude
environment provide energetic constraints upon resources available to both the
mother and the growing fetus, thereby increasing the metabolic costs of pregnancy
and serving to limit intrauterine growth. Genetic factors have been hypothesized to
influence an individual mothers physiological response to high altitude hypoxia.
Similarly, the sociocultural environment affects reproductive outcomes through the
structuring of mating possibilities and influencing demographic factors that affect,
in turn, reproductive decisions. Cultural boundaries segregate the distribution of
possible matings as demonstrated by the ways in which ethnic surnames recombine
in the next generation. Cultural influences on mating thus limit the genetic
admixture available in subsequent generations. Finally, these environmental
influences must be viewed through the lens of evolutionary history, given that
pregnancy and reproduction are integral to natural selection.
This thesis will also review the theoretical literature in order to present an
evolutionary scenario for differences that might be expected to exist in the effects of
maternally and paternally transmitted ethnic admixture on birth weight. Here, the
differential reproductive strategies of males and females are viewed through the lens
of life-history theory, based on models of optimality and resource allocation. Life-
history theory provides an understanding of how evolutionary history and the

current physical and sociocultural environments frame the allocation of metabolic
resources at an individual level. Energetically, individuals of reproductive age must
balance somatic with reproductive effort, a choice that has been shown to be highly
tuned to environmental conditions. The choice to allocate resources toward
reproductive effort is further complicated by the need to divide effort between
mating and parenting, as well as between present and future reproduction.
Life history theory predicts that the evolutionary pressures on mothers will be
different than those acting on fathers as a result of their different resource
requirements for reproduction. Resource allocation between mother and offspring is
further dictated by genetic factors that have been conceptualized as pitting the
maternal genomic interests against the paternal within the fetus. The phenomenon of
genomic imprinting operates through the parent-specific expression of alleles
affecting fetal growth. This molecular process affects fetal supply and demand for
nutrient exchange across the placenta and thereby influences fetal growth. David
Haigs conflict theory of genomic imprinting provides an evolutionary explanation
for resource competition between fetal, paternal and maternal genomic interests
(Haig 1993). Haigs model has been tested and supported with evidence from
placental laboratory animals, but only to a limited extent in humans.
Correlational population level data have not yet been utilized to examine genomic
imprinting; a sub aim of this paper is to assess the effectiveness of this approach.
The high altitude environment of La Paz, Bolivia provides a unique opportunity to
assess parent-specific contributions to birth weight for an understanding of
evolutionary and genetic factors involved. Specifically, population level data allow
us to identify the effects of variable amounts of ethnic admixture, as well as to
determine whether those effects differ by parental lineages.

In order to review the theoretical foundations for this thesis, the following
background section will present first the population factors involved in assessing
human adaptation. This will include: the relationship between ethnicity and
genetics, the differences in high altitude adaptation by duration of residence, and the
role of the social environment. The second section of the background will consider
the individual factors structuring reproductive outcomes, including: the difference in
male and female life history strategies, the role of genomic imprinting for
influencing the intergenerational allocation of resources, and evolutionary
explanations based upon competition between male and female strategies.

Population differences based upon ethnicity and genetics
Human population differences in environmental adaptation can be segmented into
those resulting from biology, from culture, and from the interaction between them.
Biological differences between populations are the result of genetic, developmental,
or environmental effects. Cultural practices can influence behavior, which in turn
can influence exposure to or the nature of biological processes. This thesis seeks to
understand, specifically, how genetic factors influence birth weight at high altitude
independently of, but also in interaction with, cultural differences. The high altitude
environment is unique compared with many other environments in that few cultural
means are available for ameliorating its primary stressor (hypoxia). However,
cultural factors can still affect reproductive outcomes at high altitude by structuring
genetic variation. In attempting to integrate cultural and biological factors affecting
reproductive outcomes, this thesis will describe how population differences affect
adaptation to the high altitude environment.
Studies of human population diversity must often address the anthropological
understanding of race if they are to evaluate differences between human

population groups. Race is the traditional taxonomic term given to describe a
genetically distinct subspecies. Historically, race in anthropology was used to
group together morphologically-distinct peoples from different geographic regions
(Molnar 1998). However, genetic data have provided us with the more precise
realization that the human species is relatively homogenous when compared to other
primates (Lewontin 1972). This homogeneity is likely the result of our recent
population explosion and extensive gene flow. Further, greater variation exists
within human population groups than between, and attempts to classify discrete
subpopulations based upon perceived discontinuous racial traits have no validity
(Templeton 1999). As a result of these data, the use of human race as a typology
lacks biological relevance (Cartmill 1999).
Two factors important to this thesis hold relevance for establishing population
distinctions. First, local populations can display differences in specific biological
traits as a result of their unique evolutionary history. Second, ethnicity, defined as
shared cultural identity, is useful for describing the differentiation of population
groups based upon social criteria. Considering population as the appropriate unit of
analysis, rather than the heuristically misleading category of race, allows us to
evaluate more detailed differences and the evolutionary factors involved. Before
addressing how natural selection may have affected high altitude adaptation, the
following considers how ethnicity can help to understand biological differences
between populations.
Ethnicity, which segments populations by cultural criteria, can be estimated in the
Bolivian population through the analysis of parental surnames. Heredity surnames
naturally contain information about relatedness within populations, as they are, like
DNA, passed down from generation to generation. Although some names are
dropped through naming conventions according to the specific kinship system, as

cultural markers surnames nonetheless contain significant information about
biological distinctions. Chakraborty has shown the utility of surnames for
delineating genetically distinct groups among the Aymara of Chile and Bolivia. He
demonstrated a reasonable agreement of ethnic classification of individuals by
name and phenotype data [as that provided by] genetic markers, (Chakraborty
1989:159) such that Aymara-named groups were predominantly (89%) Amerindian
in their genetic profiles.
Population differences thus can reflect differences in culture (ethnicity), biology
(genetics), or both. These differences can coincide with specific genetic markers and
physiological traits. However, it is important to recognize that identifying the
genetic basis for physiological differences rests within a broader social, cultural,
and political context (Atkin 2003:91).While ethnicity can be used as an estimation
for genetic factors, it also represents an important part of this [broader] context
(Atkin 2003:91) and may represent factors independent of biology. Even
considering these limitations, Chakrabortys work supports the use of ethnicity to
measure genetic admixture in our study.
The high-altitude environment
The highest of any of the worlds capitals, La Paz, Bolivia sits on the edge of the
Altiplano at a range of altitudes from 3100 to 4100 meters. This high altitude
environment presents a unique adaptive challenge for human reproduction. This is
primarily a result of the hypoxic stress induced by its low ambient oxygen pressures,
but also because of the low humidity, extreme diurnal temperature variation, and

shortened growing season (Moore 2001). Oxygen pressures decrease with
increasing altitude and, especially above 8,000 feet, hypoxia significantly affects
human physiology. As mentioned previously, studies of high altitude hypoxic stress
are especially informative about human biological adaptation because of the
difficulty of cultural mediation. While culture can influence, to a certain degree, the
effects of hypoxia, it cannot ultimately eliminate them. Thus, the high altitude
environment can be seen as a natural laboratory, where human physiology can be
observed and compared without many of the confounding factors of culture (Moore
Because low oxygen pressure imposes a significant stress upon human physiology,
studies of adaptation to hypoxia have focused upon the oxygen transport system and
the extent to which it can compensate for lower oxygen availability (Moore 2001).
Adaptations are features of structure and/or function that allow organisms to survive
and reproduce (Moore 2001). Importantly, for attributes of physiology to be
considered adaptive, they must be linked to functional outcomes. Functional
adaptations to high altitude will be those that increase the amount of oxygen that
reaches bodily tissues, thus offsetting the stress of hypoxia (if only partially) and
allowing for physiological capacities similar to those at sea level.
Pregnancy presents a unique point in the life cycle through which to view high-
altitude adaptation for two reasons: one, pregnancy is a metabolically stressful
encounter for the mother even in the absence of hypoxia and two, because events
surrounding pregnancy can exert strong influences on natural selection. Any stressor
that directly affects pregnancy and birth outcomes, including perinatal survival, will
exert a heavy selection pressure (Moore 2001).

At high altitude, a pregnant mothers energy balance is limited by the low oxygen
pressures (Mawson et al 2000). The mother provides vital nutrients for the fetus,
primarily glucose and the oxygen with which to metabolize glucose and amino acids
efficiently. Because of the vital role of oxygen in meeting fetal metabolic needs,
hypoxia presents an energetic challenge both for the mother and the growing fetus.
This sometimes results in intrauterine growth restriction (IUGR).
IUGR, defined conventionally as less than the 10th percentile of birth weight for
gestational age and sex, refers to reduction in birth weight due to a lower growth
rate. Previous studies have shown that IUGR at high altitude is due to a slowed
intrauterine growth rate rather than shortened gestation and leads to a 100 g
decrease in birth weight for each 1000m altitude gain (Jensen and Moore 1997).
Further, the contribution of IUGR to altitude-associated reduction in birth weight
has also been shown to be independent of other risk factors for low birth weight
such as smoking and diabetes (Jensen and Moore 1997).
Compared with more recent populations, long term high altitude residents display
less IUGR, possibly as a result of being able to maintain higher uterine artery blood
flow and thus greater nutrient delivery to the developing fetus (Moore et al 1998).
This proves important when considering evolutionary mechanisms, because it
potentially links maternal physiological responses to pregnancy to reproductive
fitness. As mentioned in the introduction, birth weight is an important predictor of
infant mortality. In part because birth weight is lower, infant mortality at high
altitude is greater than at low altitude. Although the selection pressure against lower
birth weight is difficult to gauge (Moore 2001), it appears likely that reducing IUGR
at high altitude improves survival and, thus, reproductive success for long-term
high-altitude populations.

The evidence for multigenerational adaptation has been displayed in comparative
studies of some of the 140 million high altitude dwellers and migrants around the
world (e.g., Himalayas, Andes, Rocky Mountains). Comparison of these populations
exhibits natural selection in action, as adaptation increases over generational time
(Moore et al 1998). Figure 2.1 displays the relationship between high altitude and
birth weight, exhibiting higher birth weights at a given altitude for the longest term
residents. Our work will show that the relationship between adaptation and length of
high altitude residence is also evident in admixted populations, which display
intermediate levels of birth weight reductions. Specifically, in our study Andeans
display the highest birth weight, Mestizos intermediate, and foreign-bom mothers
the lowest.
Figure 2.1Birth weight by altitude for different ethnic groups
(Moore 2001)
Based upon the direct relationship between generational time and birth weight and
the results of heritability studies, functional adaptations to high altitude are likely to
have a genetic basis. The results of comparative studies show that ethnically
distinct, long-term high altitude populations have adapted to hypoxia in ways that

permit them to reduce IUGR. Further, heritability studies have determined that in
many populations, a significant portion of the physiological variation in traits
affecting high altitude adaptation results from genetic variation (Beall 2000). These
data, combined with the physiological data, suggest that polymorphic variation in
the alleles influencing nutrient transfer to the fetus could have been acted upon by
natural selection.
The social environment
The social environment can structure many behavioral and physiological
interactions. Sociodemographic and cultural factors interact to affect reproductive
behaviors, including the types and number of matings that occur. This in turn limits
the amount of admixture that can occur, and thereby limits the possible genetic
heterogeneity in the offspring generation.
Assortative mating refers to the propensity for individuals to marry selected
individuals in their population. Also known as social (or status) homogamy,
assortative mating operates to provide cultural barriers to reproductive events.
Kalmijn describes two micro-level hypotheses about status homogamy: the
cultural matching hypothesis and the economic competition hypothesis (Kalmijn
1994). The cultural matching hypothesis refers to the idea that people prefer to
marry someone of similar cultural status, whereas the economic competition
hypothesis indicates that people prefer to marry someone of higher economic status.
Based upon studies in Europe, Kalmijn found that assortative mating by cultural

status is more important than assortative mating by economic status (Kalmijn
1994:422). Assortative or nonrandom mating, while largely cultural in origin, can
affect the genetic consequences of reproduction by reducing the admixture of
dissimilar individuals.
Demographic factors can affect reproductive decisions by influencing the tradeoffs
between the quality and quantity of offspring and affecting the age structure of the
population. The demographic transition refers to the move from a population
characterized by high mortality and fertility to one characterized by low mortality
and low fertility. Traditionally, the mortality rate falls initially, population growth
increases, and the fertility rate falls later in response. This transition has
occasionally been considered from a life history perspective, which is consistent
with the fact that fertility tends to lag and fall after the mortality rate has fallen. This
indicates that in less secure environments, individuals choose to have more
offspring to offset higher mortality.
Bolivia has the highest infant mortality rate (67 deaths/1,000 live births) in the
Western Hemisphere and a maternal mortality rate (390 deaths/100,000 live births)
second only to Haiti (PAHO 2001). The total fertility rate is high as well, recently
climbing to 4.0, which indicates that the population is still at a very early phase in
the demographic transition. Further, Bolivias young population indicates that a
large number of women are reaching reproductive age.
As Bolivia is the poorest country in South America, poverty is also likely to be
affecting reproductive outcomes. Nearly a third of the population is below the
international poverty line. A disproportionate amount of the poor are from the
indigenous populations. Associated with this poverty, nutritional deficits are
common, especially in the Altiplano. Caloric deficits likely play a role in Bolivias

high maternal and infant mortality rates, and also contribute to a high rate of child
malnutrition (PAHO 2001). Because a larger portion of these constraints are felt by
indigenous populations, these mothers are more likely to be energetically stressed
by social factors than are mothers with greater resources.
These social factors can indirectly influence reproductive outcomes by affecting the
types of matings that occur, the tradeoffs involved in having children, and the
resources available for raising healthy offspring. Importantly, population-level
social factors need to be viewed as interacting with the biological processes
involved in population adaptation to high altitude to fully understand birth weight
, Life history theory
We turn now from broader population factors that can affect birth weight to describe
some of the evolutionary factors that may account for differences in the maternal
and paternal effects upon birth weight. Life history theory describes how the
behavior and physiology of organisms are guided by natural selection. It is pertinent
to considerations of birth weight insofar as it emphasizes the various strategies
available for optimal resource allocation. In essence, phenotypes are judged by
their consequences for reproductive success. Accordingly, organisms must balance
the complexities of survival, growth, development, and reproduction with the
energetic constraints of their unique environment if they are to leave descendents
(Chamov 1993, Hill and Kaplan 1999, Tracer 2002). Life history theory predicts
that the allocation of energy to these various processes, as well as the specific

timing of life events, results from physiological tradeoffs that have, as their end
goal, optimizing reproductive success.
Life-history theory conventionally segments the allocation of energy budgets into
somatic and reproductive effortwith somatic effort referring to basic
maintenance, growth and development; and reproductive effort referring to the
metabolic energy needed for reproduction. Reproductive effort can be further
divided into mating and parenting effort. Reproductive strategies, in turn, are
conceived as suites of coevolved anatomical, physiological, and psychological
traits designed by natural selection for the optimal allocation of mating and
parenting effort (Chishholm 1993:2).
While in energy-abundant environments, it might be entirely reasonable for the
principle of allocation to be nearly violable (Tracer 2002:622) such that energy
trade-offs are not as costly, in energy-restricted environments like high altitude, life
history trade-offs become more pronounced. In these environments, the fitness costs
and benefits corresponding to reproductive decisions closely dictate how resources
are allocated. Organisms must decide, based upon knowledge of the environment,
which strategy will offer the greatest fitness benefit. As a broad generalization, for
organisms in unpredictable environments where offspring survival is less secure, it
often proves more beneficial to spread reproductive efforts among many offspring.
In a more predictable environment, greater parenting effort can be invested in each
offspring, thereby increasing their intergenerational fitness. Importantly, life history
theory leads to the expectations that there exist a variety of optimal strategies for
increasing fitness. Two fundamental concerns drive these strategies: first, the
maximization of ones own lifetime net energy production for reproduction; and
second, the maximization of total offspring energy production (Hill and Kaplan

The idea of embodied capital provides an effective heuristic device for
understanding the intergenerational allocation of resources. In this model, developed
by Hillard Kaplan, individuals can invest not only in capital embodied in their own
soma, but also in the capital embodied in offspring. However, such allocations
decrease resources available for the production of other offspring and, hence,
decrease the total quantity of offspring produced (Hill and Kaplan 1999:402). The
concept of embodied capital is useful in understanding the optimal allocation of
resources among parents and their offspring. In many energy-restricted
environments, this allocation becomes quite acute, such that the ultimate concern
becomes one of parental investment versus parent survival (Tracer 2002).
The partitioning of energy to reproduction, specifically to pregnancy, sits at the
cornerstone of an effective life-history strategy. Mediation of reproductive effort
can be thought of as occurring at two primary levels, the behavioral and the
physiological (Hill and Kaplan 1999). Importantly, both reproductive behavior and
physiology differ significantly for males and females. Reproductive ecology has
developed into a robust field of inquiry concerning the different reproductive
behavioral strategies of males and females, as well as into how those strategies
respond to or map onto the ecological terrain (Chamov 1993). The effectiveness
of various reproductive behaviors in terms of fitness consequences depends upon
the tradeoffs provided by the unique socioecological context.
The difference in male and female strategies in both behavioral, as well as
physiological approaches to reproduction, results from the simple fact that females,
as carriers of the offspring, bear the greater metabolic burden. Indeed the female
reproductive strategy is ultimately the rate limiting step in primate reproduction
(Chisholm 1993). Compared to other primates, humans have a relatively long

gestational and weaning period. Females must bear the fetus through gestation,
provide for lactation, and usually are the greater source for parental investment
afterward. Males, in contrast, often optimize their reproductive success by diverting
more energy to the mating end of the mating-parenting continuum (Hill and Kaplan
1999). Perhaps predictably, polygynous societies, which fit both male and female
needs (where males maximize mate quantity and females maximize mate quality),
are the most common primate mating systems.
Physiologically, the endocrine system controls metabolic allocation between
somatic, parenting, and mating effort. Because of the greater metabolic burden
present throughout pregnancy, a females reproductive physiology is far more
sensitive to environmental energy constraints than a males (Ellison, 2003).
Although testosterone levels vary considerably in influencing energy allocation
between mating and parenting effort, male gametic production is often quite
constant even in scarce environments (Ellison, 2003). While males do not spend as
much metabolic energy in reproduction, females must devote a greater portion of
their somatic energy towards it. Predictably, female fecundity, specifically
ovulation, is sensitive to environmental stressors (Ellison 1990). Females, in
investing in any given offspring, utilize resources at a cost to their own somatic
requirements, as well as to the detriment of their future ability to procreate. Parental
investment in any one offspring, likewise, must naturally be spread among greater
numbers of offspring as parity rises.
The cost of reproduction for mothers is illustrated by the inverse relationship
between maternal energy stores and parity (Tracer 2002). Females who reproduce
frequently and do not maintain large interbirth intervals are subject to maternal
depletion syndrome (Ellison 2001), in which higher parity is associated with
reduced fat stores. Undernourished or overworked women often have further

depletion in their fat reserves, which reduces their ability to invest as much in their
offspring during gestation and leads to low birth weight, premature birth, and
increased risk of infant mortality (Ellison 2001). Perhaps as an example of this
reduced ability to procure reproductive energy, birth weights drop in some cases as
parity increases above a certain threshold (in many cases, as in our sample, this
threshold appears to be four live births) (Tracer, Personal communication).
The duration of gestation and birth weight are, correspondingly, both sensitive to
maternal energy availability, as restricted energy availability can have a
particularly negative effect on fetal fat accumulation and resulting birth weight
(Ellison 2001:72). Lower maternal energy balance can reduce birth weight as the
result of either shorter gestation or intrauterine growth restriction. Low energy
availability is associated with shorter gestational duration (Kline et al 1989), and
gestation is somewhat shorter in the energy-restricted high altitude environment
(Jensen and Moore 1997).Variations in fetal fat accumulation account for a large
amount of this variation in overall birth weight. While fetal fat makes up only about
14% of newborn birth weight, in some studies it accounts for up to 47% of the
variation in birth weight (Haig 1993).
Pregnancy can also be viewed as a metabolic conflict between fetus and mother.
The mother must metabolize enough nutrients for herself and the fetus, and fetal
growth is especially energy intensive. The fetal growth rate increases throughout
gestation and peaks at about 36 weeks, when a large amount of fat stores are laid
down (Haig 1993). After that point, the fetal growth rate slows remarkably until
birth. If gestation continues until the 40th week, the fetus must often draw upon its
own fat stores. The metabolic crossover hypothesis (Ellison 2001) describes this
transition from maternal sources of energy to fetal sources. The fetal metabolism is
able to increase until it reaches the maximal level of metabolic resources that the

mothers physiology can provide, as she cannot provide enough nutrients to meet
fetal metabolic needs past this level. The fetal growth rate slows as the relative
maternal investments in fetal growth diminish. Any further needs are increasingly
drawn upon from the fetuss own fat stores. This has led to the idea that birth occurs
when the fetus begins to starve in-utero, or that point at which it would be able to
acquire greater nutrients outside of the womb. Once outside the womb the fetus is
able to procure greater amounts of necessary fat stores through maternal lactation
than placentally-transmitted nutrients. The decreased oxygen availability at high
altitude further reduces the mothers ability to provide adequate nutrients for the
fetal metabolism, which may partially explain the slightly shorter gestational
periods observed in some studies at high altitude (Jensen and Moore 1997; Gonzales
etal 1993).
The metabolic situation of pregnancy has often been conceptualized as a
cooperative one between the mother and fetus, as both are invested in fetal survival.
However, as the metabolic crossover hypothesis attests, maternal and fetal needs
sometimes conflict with each other. With this in mind, we turn from the
understanding of energetic resource allocation provided by life-history theory,
which focuses primarily upon the parent, to this idea of reproduction as a conflict
between parent and offspring.
Parent-offspring conflict
Robert Trivers conceptualized the notion of parental investment in order to
describe the relationship of a parents investment in its offspring to reproductive

success (Trivers 2002). This relationship hinges upon Hamiltons earlier idea of
inclusive fitness, which provides the foundation for the genetic basis of Trivers
argument (Hamilton 1964). Inclusive fitness refers to the fact that fitness is not just
a function of an individuals reproductive success, but rather the reproductive
success of any of the individuals genetic relatives. Robert Trivers parent-offspring
conflict model utilizes this understanding, but provides a conceptual shift from a
focus on the parents investment to a perspective whereby the offspring is an actor
in the interaction (Trivers 2002). Accordingly, conflict must be assumed to lie at
the heart of sexual reproduction itselfan offspring attempting from the very
beginning to maximize its reproductive success would presumably want more
investment than the parent is selected to give (Trivers 2002:129). Further, this
relationship is closely balanced by the fact that, based upon inclusive fitness, the
parent, offspring, and the offsprings siblings share a close, but unequal genetic
relationship. This balance prevents the allocation of resources from swinging too far
in an individual offsprings direction, but provides enough of a difference that
parent and offspring can often have competing goals.
These competing goals, produced by a tenuous balance between a parent and an
offspring who share some, but not all, of each others genes, led David Haig to posit
that pregnancy is ultimately an intergenerational genetic conflict. To do this, Haig
applied Trivers notion of genetic conflict to the somewhat enigmatic phenomenon
of genomic imprinting. Haigs conflict theory has provided an elegant framework
for explaining much of the imprinting data, using the perspective of competing
evolutionary drives (Haig 1993). Before returning to the more theoretical
conceptions of conflict theory, the following section describes the mechanisms of
genomic imprinting.

Genomic imprinting
Genomic imprinting refers to the differential expression of a gene (actually an
allele) based upon the parent from which the allele originated. This parent of
origin effect leads to tissue-specific nonexpression (imprint) of the allele at some
period of development (Spencer et al 1999). This hemizygotic expression has
confused some researchers, because by leaving the individual as a functional
haploid for a given locus, it would seem to offset the benefit of sexual reproduction
and leave the individual vulnerable to deleterious mutations (Spencer et al 1999).
Several theories, which will be clarified shortly, attempt to explain a selective
advantage of uniparental expression that outweighs this cost.
DNA methylation appears to be the key molecular mechanism driving imprinting. It
marks the imprinted genes differently in egg and sperm, and the inheritance of
those epigenetic marks leads to differential gene expression (Reik and Surani
1997:1). Gene coding regions that are methylated cannot be transcribed. While
somatic imprints are maintained throughout an organisms life cycle, imprints of
germ cells are erased at an early developmental stage in the new organism, but
reimprinting occurs at a later stage when germ cells generated.
About 60 imprinted genes have been identified in the mouse genome, and about 45
of these are conserved in humans. The phylogenetic distribution of imprinting
indicates that it occurs rarely in invertebrates and many other organisms, with the
bulk of the examples found in mammals and angiosperms (Reik and Surani 1997).
Curiously, while non-mammalian vertebrates often have extensive methylation, they

do not show much imprinting. Non-mammalian vertebrates, such as those that lay
eggs, lack the intimate contact and sharing of resources between parent and
offspring. Because imprinting has only been shown to occur in organisms with
intimate contact of parent and offspring cells during development, the phylogenetic
distribution has supported the notion that imprinting plays a role in resource control
across a parent-offspring barrier, such as the mammalian placenta or angiosperm
endosperm barrier.
Most imprinted genes influence perinatal growth; some have even been shown to
affect postnatal behavior. The growth effects of imprinted genes, together with their
phylogenetic distribution, have supported the notion that imprinting is thought to
influence the transfer of nutrients to the fetus and the newborn from the mother
(Reik et al 2003:). Interestingly, many of the imprinted genes affect fetal growth in
an antagonistic manner. Specifically, genes that are paternally expressed tend to be
growth promoters, whereas those that are maternally expressed tend to be growth
suppressors (Reik and Surani 1997). For example, while insulin-like growth factor
II (IGF2) is paternally expressed in both fetal and placental tissues, the regulation of
the insulin-like growth factor II receptor, which degrades IGF2, is often maternally
controlled (Figure 2.2) (Mochizuki 1996). The parent-specific regulation of growth
factors has been displayed through the discovery of uniparental disomies, which
result from receiving both alleles at a locus from one parent. Studies were initially
fairly conclusive in showing that unipatemal disomies result in larger than average
progeny, and maternal disomies the opposite (Reik and Surani 1997). However, the
data on disomies are now a bit more ambiguous, because copies of paternal alleles
do not uniformly lead to larger offspring, and copies of maternal alleles do not
always result in smaller offspring.

Figure 2.2 Some paternal and maternal growth factors act in direct opposition at the
cellular level. This figure shows an example of how paternal and maternal
expression of growth factors conflict: paternally produced IGF-II has its growth-
promoting effects via the type I receptor, but is degraded via the maternally
produced type II receptor (Haig 1999).
Many imprinted genes occur in clusters on chromosomes, indicating that a small
number of genes may operate as a functional unit to have wide-ranging
physiological effects. Table 2.1 provides a list of imprinted genes in humans, along
with their chromosomal location and physiological function.

Table 2.1Known imprinted genes in humans, their location and function
Gene Function Location Active allele Chrom. #
IGF2 Growth factor Labyrinthine trophoblast Paternal 11
INS2 Growth factor Paternal 11
HASH2 Transcription factor Extravillous trophoblast Maternal 11
IGF2R IGF2 antagonist Maternal 6
INS1 Growth factor Pancreas, thymus Paternal 11
PEG3 Transcription factor Villous cytotrophoblast Paternal 19
MEST/PEG1 Transcription factor Villous mesenchyme & trophoblast Paternal 7
H19 Untranslated RNA Maternal 11
Evolutionary explanations
The phenomenon of genomic imprinting provides a perspective on maternal-fetal
conflict that is both molecular and evolutionary. Several theories have attempted to
explain the evolution of imprinting, but none has captured as much attention and
acceptance as David Haigs conflict theory.
Haigs theory of genetic conflict applies parental investment theory to maternal-fetal
interactions in pregnancy. Specifically, he utilizes the idea of optimal resource
allocation from life-history theory, along with Hamiltons inclusive fitness and
Trivers parent-offspring conflict theory to help understand the genomic conflict

between mother, father, and offspring. Two related conflicts occur as part of Haigs
main theory: that between the paternal genome and the maternal genome, and that
between the offspring and mother. As in parent-offspring conflict, the conflict
theory of genomic imprinting depicts the conflict over maternal resources between
the mother and the growing fetus at the genomic level. Because the fetus obtains
half of its genome from the father, its genetic interests do not perfectly match the
mothers (Haig 1993).
Haigs theory provides in essence a reductionistic view centering upon the operation
of alleles whose effects differ, depending upon the parent of origin. In other words,
the effect of a maternally-derived allele on matrilineal kin differs from the effect of
a paternally-derived allele on patrilineal kin. Once genomic imprinting arises,
natural selection will favor alleles whose effects differ depending upon the alleles
parent of origin (Haig 1993). The evolutionary origins of and mechanisms
responsible for imprinting are more difficult to understand, but some viruses
methylate regions of DNA; it is hypothesized that imprinting could have originated
from a virus and then rapidly developed the parental biases.
Accordingly, Haigs hypothesis argues that parent-of-origin effects are the outcome
of conflicting selective forces between maternal and paternal lineages (Haig 1993).
This conflict requires the presence of asymmetric kin, i.e., those who do not share
identical patrilineal and matrilineal fitness interests. While full siblings have
symmetric interests, half-siblings with different fathers will be asymmetric
because they do not share the same patrilineal fitness. Because of this kinship effect,
the conflict theory has also sometimes been dubbed the kinship theory (Haig
2000). Through the use of game theory, Haigs theory predicts that evolutionarily
stable strategies will evolve such that the effects of alleles at a given locus are
balanced. These strategies will either be symmetric or asymmetric. Symmetric

strategies will occur where matrilineal effects and patrilineal effects favor the same
level of gene expression. Asymmetric strategies will occur when the lineages favor
different total levels of gene expression (Haig 2000). With the advent of genomic
imprinting, selection will tend to favor the selection of the dominant allele and
complete imprint of the other.
Kinship asymmetry will occur as a result of even fractional amounts of multiple
paternity, because in this situation all siblings are not related to one another in a
symmetric fashion (Spencer et al 2000). Humans tend to have less multiple paternity
than other species, but population genetic models indicate that only a small amount
is enough to allow genomic imprinting to evolve as a stable strategy (Spencer 2000).
In the case of multiple paternity, while a mother will be equally related to all of her
offspring, a father cannot be sure that each of the mothers offspring share his genes.
Therefore, from an evolutionary vantage point, it is advantageous for the fathers
genes to require an increase in maternal investment in the offspring. Alleles at
imprinted loci will thus tend to accumulate parentally antagonistic effects, such that
paternally expressed loci will make greater demands upon mothers (Haig 1993).
Whereas the mother must balance resources between present and future
reproduction, a fetus carrying the paternal genome will desire to capture as many
resources as it can. The expected outcome here would be for paternally-derived
alleles to have been selected for creating larger offspring with improved chance of
survival. The mother, in contrast, will aim for greater balance between any given
pregnancy and resources required for future pregnancies.
Two other theories explain much of the accumulated data on genomic imprinting:
the ovarian time bomb theory and the variance reduction hypothesis. The
ovarian time bomb theory does not require intergenerational conflict. Rather, this
theory posits that genomic imprinting evolved to prevent parthenogenesis (Reik and

Surani 1997), or the development of an unfertilized egg. The spontaneous
development of an unfertilized egg can lead to invasive trophoblast disease, where
developing tissues invade the uterine wall, and place huge selective costs upon
mothers. In a recent article in Nature, Kono et al attempted to breed parthenogenetic
mice, but this proved difficult due to the presence of genomic imprinting. However,
under conditions of elevated production of placental growth factor they were able to
induce parthenogenesis but the embryos died shortly thereafter, indicating that the
paternal genome was necessary for successful long-term completion of pregnancy
(Kono et al 2004). These results are consistent with the ovarian time bomb
hypothesis. However, the ovarian time bomb hypothesis does not argue for a
conflict in growth factors or the direction of parental growth effects predicted by
Haigs theory. Rather, the prevention of parthenogenesis appears to be a cooperative
event between parents.
The variance reduction hypothesis is more consistent, on the whole, with the
accumulated data on imprinting. This hypothesis asserts that genomic imprinting
evolved to reduce the variation in birth weight and for a fine tuning of birth weight
to maternal and environmental conditions (Reik and Surani 1997). As a fine-tuning
mechanism, imprinting could potentially allow for greater flexibility in response to
changing environmental demands. The variance reduction hypothesis has
considerable support, the major exception being that it posits a cooperation between
maternal and paternal interests. The oppositional effects of paternal and maternal
growth factors observed previously (Fig 2.1) in many studies argue against this
hypothesis, but support the conflict theory.
These three theoriesconflict, ovarian time bomb, variance reduction,have
been compared on their ability to explain the phylogenetic distribution of genomic
imprinting, the fact that imprinted genes occur in clusters on chromosomes that

affect fetal and placental growth, and the oppositional effects of paternal and
maternal growth factors. Primarily because the ovarian time bomb and variance
reduction theories cannot adequately explain the specific direction (i.e., paternal
factors increasing growth, maternal reducing) of these oppositional growth effects in
fetal and placental cells between paternally and maternally expressed genes, Haigs
conflict model provides the best explanation.
Genomic imprinting and placental physiology
Genomic imprinting is a potentially important mechanism in placental mammals
specifically because it is thought to influence the transfer of nutrients to the fetus
and the newborn from the mother (Reik and Walter 2001:21). As a corollary to the
conflict theory, the supply and demand theory postulates that imprinted genes in
the placenta control the supply of nutrients, whereas in fetal compartments they
control nutrient demand by regulating the growth rate of fetal tissues (Reik et al
2003). Trophoblast cells are fetally-derived cells in the placenta. Paternally-
expressed gene transcripts are found in the labyrinthine trophoblast and
spongiotrophoblast of the placenta, including the glycogen cells, and in
labyrinthine blood vessels (Mochizuki et al 1996). The invasiveness of placental
tissues is positively correlated with the relative expression of paternally-derived
chromosomes (Reik and Walter 2001). Imprinted genes can affect either the overall
growth of the placenta or of the particular structures that determine specific
transport systems.

The hemochorial placenta-shared by rabbits, moles, rodents, and higher primates
is unique in that it has a very large surface area for nutrient transfer. Greater
placental size leads to greater surface area, which allows for greater nutrient
diffusing capacity. The placental layer functions in the promotion of fetal growth
and viability by mediating interactions between the mother and fetus through its
feto-matemal interfaces (Reik et al 2003:39). Genomic imprinting has been shown
to regulate the spatial interactions between different cell types within several types
of feto-matemal interactions, which can affect placental diffusing capacity; genomic
imprinting also plays a role in controlling the growth of these cells.
In organisms with a hemochorial placenta, materials pass between maternal and
fetal bloodstreams through a single vessel wall. Further, the chorionic villi of the
fetal circulatory system lie in the intervillous space. This anatomy allows the fetus
to gain direct access to circulating maternal blood. Figure 2.3 depicts a simple
model of maternal blood flow through the placenta. This anatomy requires the
maintenance of high concentration gradients of key nutrients and gases as well as a
high blood pressure to maintain circulation (Ellison 2001). The two major energy
molecules in circulation, oxygen and glucose, are especially important for fetal
growth. Glucose is the primary fuel source for the growing fetus, and oxygen is
required to bum glucose into metabolizable form (Ellison 2001). Preeclampsia,
characterized by high blood pressure and protein in the urine, could be interpreted as
the maternal consequence of insufficient uteroplacental supply to meet fetal
demands for these nutrients.
The placenta is able to release hormones into the maternal circulation, which can
then manipulate maternal physiology for fetal benefit (Haig 1993). Haig terms
placental hormones that act upon maternal receptors as allocrine hormones, to
emphasize that they are produced by one organism to act on the receptors of

another (Haig 1993:506). Further, during implantation, fetally-derived trophoblast
cells invade the maternal endometrium and remodel the endometrial spiral arteries
into low-resistance vessels that are unable to constrict (Haig 1993:515). This
invasion allows the fetus direct access to the mothers arterial blood and reduces
uteroplacental vascular resistance, which in turn, raises blood flow. Both factors
prevent the mother from reducing the nutrient content or blood flow to the fetus
without reducing her own. Preeclampsia is often the result of incomplete invasion.
Early studies in genomic imprinting found that placental development is influenced
by paternally-expressed alleles, insofar as offspring lacking paternal alleles
developed poorly-formed placentas. For example, instances of double paternal
expression, known as uniparental disomies, produce hydatidiform moles, which are
characterized by a grossly-large placenta and a reduced or absent fetus. Several
paternally expressed alleles (e.g., PEG3, a transcription factor) stimulate growth in
the placenta. Somewhat counter to Haigs predictions, HASH2, a maternally-
expressed gene observed in the extravillous trophoblast, controls the extent of
placental invasion.

Figure 2.3 Maternal placental circulation. This figure depicts a simple model of
maternal circulation through the placenta during pregnancy. Maternal blood is
pumped from the right side of the heart through the lungs where it is oxygenated.
Then it is pumped by the left side of the heart throughout the maternal circulation, a
portion of which flows to the uteroplacental circulation during pregnancy. (Haig
1999). This portion is a function of uteroplacental (Rp) versus maternal (Rm)
vascular resistance.
To summarize, this background has described both broad population factors and
more direct individual influences on birth weight at high altitude that are affected by
evolutionary forces. Broad population factors can influence the admixture of genes
that may have been selected for and therefore could facilitate adaptation to
pregnancy at high altitude. The actions of such genes may be expected to affect an
increase in uterine artery blood flow. Other genetic factors could interact with those
influencing uterine artery blood flow by, for example, affecting fetal growth at high
altitude. Maternal genetic factors could affect blood flow, via the process of
genomic imprinting, or through an interaction between the two. Paternal genetic
factors are also involved in the process of genomic imprinting, and may influence,
in turn, the resource transfer from mother to fetus through the placenta. Thus, as
portrayed here, both the population and the individual perspective emphasize the
role of evolutionary processes in shaping resource allocation.

Data collection and definition
Data for this study were collected as part of a larger NIH-funded project titled
Interpopulational Differences in Intrauterine Growth Restriction at High Altitude,
conducted by Loma Moore to determine the mechanisms by which populations with
a long duration of high-altitude ancestry are protected from altitude-associated
IUGR compared with persons of low-altitude ancestry living at the same altitudes.
As part of her study, a retrospective review of medical records was conducted in
four cities in Bolivia (Santa Cruz, Cochabamba, Oruro, and La Paz) at altitudes
ranging from 300 meters to over 3,800 meters. Only the data from La Paz, at an
altitude of 3,600 meters, are utilized in this study.
Charts were identified for all births between January 1998 and April 1998 to insured
persons at the CajaNacional de Salud (CNS, or National Health Care Fund) and
from January 1996 to April 1999 at a private clinic serving wealthy Bolivians and
foreign individuals. We specifically examined birth weight data and the parents
surnames in order to ascertain the effect of parental ethnicity on birth weight. Data

were included for 1385 singleton births. For inclusion, cases were required to have
birth weight, gestational age, and at least 1 surname each from the father and the
mother. Extreme outliers for birth weight (n=20), all of which had a birth weights
less than 1000 grams and a gestational age less than or equal to 32 weeks, were
removed. 11 of these were intrauterine deaths.
In the Hispanic system for assigning surnames, women do not change their surname
when married and each parent retains the name of both his/her parents (Fig 3.1).
Population ancestry was determined by inspection of the two maternal and two
paternal surnames for each birth, a method which has been shown to correlate well
with genetic marker-based assessment of population ancestry (Chakraborty 1989).
The Bolivian study team members, chiefly Dr. Enrique Vargas, classified the
surnames as to whether they were Andean (Aymara or Quechua), Foreign
(European or other non-Hispanic origin), or Mestizo (mixed) which were largely
names of Hispanic origin. Parental ethnicity was then classified into one of five
categories depending upon whether the parental surname combinations were
Andean-Andean (AA), Andean-Mestizo (AM), Mestizo-Mestizo (MM), Mestizo-
Foreign (MF), or Foreign-Foreign (FF). This classification is irrespective of the
order of each parents surnames; ie., the parental category AM includes both those
with an Andean father and a Mestizo mother and those with a Mestizo father and an
Andean mother. The Andean-Foreign category was excluded because that surname
combination occurred very infrequently (only 14 persons, 6 mothers and 8 fathers,
were classified as Andean-Foreign).

Figure 3.1Hispanic assignment of maternal and paternal surnames.
Table 3.1 Variables assessed Mean+sem [95%CI] Min Max
Mothers age (yr)
Maternal job (unskilled/skilled)
Maternal weight gain (%)
Maternal weight gain, absolute (kg)
Preeclampsia (no/mild/severe)
Parity (# livebirths)
Gestational Age (wk)
Birth weight (g)
Babys sex (m/f)
16 47
unskilled-45% [42.3,47.7]
skilled-55% [52.3,57.7]
14+.27 -22 74.3
8+.15 -16 39
No-82% [80,84] Mild-17% [15,19]
Severe 2.2.03
M-53% [50.3,55.7]
F-47% [44.3,49.7]
0 9
32 42.9
1310 4630

The variables obtained from the birth records included birth weight, gestational age,
parity, mothers absolute and percent weight gain, mothers age, preeclampsia
(mild/severe), and babys sex (Table 3.1). Birth weight was measured by hospital
personnel at the time of birth. Gestational age was determined by self report as
weeks from either the first day of the last menstrual period (LMP) or by clinical
estimate based upon physical exam at birth. The value obtained from LMP was used
unless it was more than two weeks discrepant with that obtained from the LMP, in
which case the value based on clinical exam was used. Parity was determined by
self report of obstetric history and included the current delivery. Preeclampsia was
defined by the presence of > 2 blood pressure (BP) readings at least six hours apart
in women which were > 140/90 mmHg or a systolic rise >30 mmHg or a diastolic
rise >15 mmHg from a 1st trimester value in women who were normotensive while
nonpregnant, and >1+ proteinuria by dipstick or >0.3 g/L protein in a 24-hr.
collection. Severe preeclampsia was defined as BP > 160/110 or when signs of
impending seizure such as blurred vision or epigastric pain were present. Maternal
employment was categorized as either manual/unskilled or office/skilled by
inspection of the nature of her employment. Socioeconomic status is often measured
based upon occupational class, education level, or income level. This paper uses
maternal occupational status to estimate socioeconomic status. Individuals with
manual/unskilled jobs are considered to be of lower socioeconomic status,
whereas individuals with office/skilled employment are considered to be of
relatively higher socioeconomic status.

Birth weight differences were analyzed based upon paternal and maternal ethnicity
using an analysis of covariance (ANCOVA) with gestational age and parity
included as covariates. The maternal and paternal effects upon birth weight were
initially assessed separately to understand their independent contribution. Main
effects for paternal and maternal ethnicity were then analyzed together in a full
effects multiple analysis of covariance (MANCOVA) with gestational age and
parity as covariates. Main effects, covariates, and interaction terms were compared
in a correlation matrix to determine if multicollinearity was present. High
collinearity (>.70) between variables supported the removal of one of those
variables from the model. Variables significant at the alpha=0.05 level were retained
in the model as covariates. All other variables were assessed independently using
one-way ANOVAs in multiple comparisons.
Linear contrasts were examined for both maternal and paternal ethnicity in the
ANCOVA and MANCOVA to assess the linearity of the relationship between
ancestry and birth weight. Finally, both maternal and paternal ethnicity were
included in a categorical regression as ordinal variables, with gestational age
(continuous) and parity (ordinal) as additional independent variables. The
categorical regression model allows for a better estimation of effect size than a
traditional linear regression model does when the independent variable is
noncontinuous (categorical or ordinal). P values in tables refer to tests of ANOVA
tests for mean difference unless otherwise specified. Statistical significance is

indicated by a p-value of less than .05, whereas a marginally significant trend
refers to a p-value between .05 and .10. Data are reported as mean + SEM (standard
error of the mean), or 95% Cl (confidence interval) for proportions. Data were
analyzed using SPSS 12.0.

Distribution of matings
The mating distribution (see Figure 4.1) displays a strong degree of asymmetry
towards social homogamy, with individuals of similar ethnicity mating more
frequently with each other than would occur randomly. The correlation coefficient
between mothers and fathers ethnicity is .556 (p<.001), illustrating the presence of
assortative mating in this population. Further, the distributions of ethnicities are not
uniform across all categories; for example, there were no matings between fully
Andean and Foreign parents. This is consistent with the combinations of surnames
present in the previous generation, insofar as Andean-Foreign only occurred in 0.5%
of the matings. This category was ultimately removed from the analysis. Most
matings (57%) occur between individuals of the same ethnic background (Andean-
Andean, Mestizo-Mestizo, or Foreign-Foreign) or between either Andean or Foreign
individuals with Mestizos (34%).
The presence of assortative mating limits the number of discordant admixture
combinations (i.e., AA-FF). This potentially reduces some of our ability to detect

differential maternal and paternal effects across all ethnic combinations. Because
our sample includes ample matings of Mestizo individuals with all other admixture
categories, we consider the Mestizo group to be the best group for analysis.
Figure 4.1Distribution of matings by maternal and paternal ethnicity
Maternal Ethnicity
A AM M MF AF F Totals:
Totals: 179 216 815 99 6 83
= 42%
= 40%
= 13%
= 2%
= <2%
This figure shows the proportions of various mating combinations, with %
expressed as % of total sample size.

Differences in birth weight between ethnic groups
Birth weight varies systematically among the various surname combinations, falling
progressively with increasing foreign admixture. This is true using either the
maternal or paternal surname for classifying ethnicity (Table 4.1 and Figure 4.2). A
parental gender effect was also present such that the lowest birth weights occur
among mothers of foreign ancestry, whereas the highest were found among fathers
of Andean ancestry. Further, birth weight classified maternally was lower than birth
weight classified paternally for all categories except for the Mestizo group.
Table 4.1 Mean birth weight values classified by parental ethnic category
Ethnicity Classified Classified
paternally maternally
BW +/- se N BW +/- se N
AA 3221.9+/-34 184 3168+/-34.5 176
AM 3183.6+/-33.2 183 3176.7+/-30 215
MM 3062+/- 15.7 844 3085 +/- 15.7 814
MF 3100+/-39.8 100 3033.6+/-46.7 97
FF 2983 +/- 55.2 74 2938+/-48.6 83
Total 3097.3 1385 3097.3 1385
p<.001 p<.001
Birth weight differs by parental ethnicity group, either when classified maternally or
paternally (p<.001). Figure 4.2 illustrates the fall in birth weight when classified by
either paternal or maternal ethnicity. However, this fall is equivalent between

maternal and paternal classifications. Neither the slopes of the lines, nor their
intercepts, are significantly different from one another.
Figure 4.2Birth weight classified by parental ethnic admixture
2 3100
55 3000
£ 2900
This figure illustrates the association of birth weight with ethnicity, when classified
either paternally or maternally. Birth weights are shown for the 1385 babies in our
study in two fashions: first by classifying by fathers ancestry and then by
classifying by mothers ancestry.

Birth weight can also be influenced by gestational age, parity, and maternal age.
When birth weight is analyzed by maternal and paternal ethnicity in an ANCOVA
model with gestational age and parity as covariates, paternal ethnicity is marginally
significant (p=056) and demonstrates a significant linear trend (p=.038) with birth
weight. Maternal ethnicity is significant (p=010) with birth weight, with a
significant linear trend (p=006). When maternal and paternal ethnicities are entered
simultaneously in a full effects MANCOVA model, with gestational age and parity
as covariates, the maternal effect is again significant (p=.044), whereas the paternal
effect is not (p=.222). The maternal linear trend is marginally significant (p=.087),
whereas the paternal linear trend is not significant (p=.295). These results indicate
that the decline in birth weight with increasing foreign admixture shows a greater
effect and a stronger linear trend for mothers than for fathers. In a categorical
regression model with paternal ethnicity, maternal ethnicity, parity (ordinal) and
gestational age (continuous) as variables, the maternal standardized beta coefficient
(-.074, p<.001) is 80% greater than the paternal coefficient (-.040, p=.065). The
categorical regression model is thus consistent with the ANCOVA models.
The Mestizo category, which includes all births with at least one Mestizo parent,
was the largest in our sample, making up 77% of the total. We performed more
intensive analysis in this group, because it has a more continuous distribution of
matings across all ethnic categories. The effects of ethnic admixture in the Mestizo
group are consistent with its effects in the entire sample, which is important because
this supports the presence of independent effects of maternal and paternal ethnic

admixture upon birth weight and allows further assessment of the difference
between the maternal and paternal effects. Figure 4.3 displays the relationship of
birth weight with maternal or paternal ethnicity when one parent is Mestizo. In this
figure, the ethnicity of the non-mestizo parent is indicated on the x-axis. In the full
effects MANCOVA, the effect of maternal ethnicity is significant (p=.038), with a
significant linear trend (p=.018), whereas paternal ethnicity is not (p=.162). Paternal
ethnicity is however marginally significant in an ANCOVA model (p=.091). In the
categorical regression, the standardized beta for maternal ethnicity (-.084, p<.001) is
nearly twice that of paternal ethnicity (-.043, p=.09). Thus, as was seen in the full
sample, the decrease in birth weight with increasing foreign parentage is greater for
maternal than paternal ancestry (Figure 4.3).
Figure 4.3Birth weight classified by ethnicity in the Mestizo group

This figure displays the association of birth weight with parental ethnicity, when
one of the parents is mestizo. Ethnicity here represents the ethnicity of the non-
mestizo parent.
Table 4.2 Mean values for covariates, categorized by maternal ethnicity
Maternal ethnicity AA AM MM MF FF
Gestational Age (wks) 39.3+. 1 39.4+. 1 38.7+.06 38.5+.2 38.9+.04 p<.001
Parity (# live births) 2.6+.1 2.4+.1 2.1+.04 2.1+.1 1.9+.14 p<.001
Maternal age (yrs.) 27.8+.4 27.7+.4 30.1+.2 30.1+.5 31.0+.6 p<.001
Gestational age
In all models, gestational age has the strongest effect on birth weight (p<.001). This
is consistent with previous findings in the literature (Jensen and Moore 1997),
which show that gestational age is directly related to birth weight. Gestational age,
like birth weight, falls with increasing foreign ancestry (Table 4.2). This likely has
two causes: first, gestational age at high altitude, like birth weight, is shorter in
relative newcomers. Second, foreign and Mestizo mothers are more likely to have
cesarian sections than their Andean counterparts and cesareans are commonly
scheduled before full term. To have cesarian sections is considered prestigious in
Bolivian society since it implies that the family has sufficient wealth to afford the

delivery and to permit the mother to be inactive following delivery. In our sample,
whereas mothers with some Andean ancestry (ethnic categories AA and AM) had
cesarian sections in only 20% of all births, Mestizos and mothers with Foreign
ancestry (MF and FF) had cesarian sections in 52% of births. Physician convenience
is probably also a factor. Nonetheless, although gestational age shows a strong
effect upon birth weight, the difference in gestational age between groups is only
about half of a week. While gestational age contributes a significant portion of the
birth weight variance, this half-week difference cannot account for the roughly 240
gram difference between babies of Andean and Foreign parentage. Jensen and
Moore (1997) estimate that a roughly 0.5-week difference like we show here
between Andean and Foreign mothers would be expected to account for only an
approximately 40 g reduction in birth weight.
Parity is also greater in Andean women than in Mestizo or Foreign women. This is
the likely result of Bolivia not having completed the demographic transition a high
infant mortality rate persists. In fact, Bolivia has one of the highest infant mortality
rates in South America, which often corresponds with a similarly high fertility rate.
Parity tends to be related to birth weight such that primiparous mothers tend to have
the smallest babies, with birth weight increasing for subsequent births, but leveling
and even declining slightly again after the fourth (Figure 4.4). This is a similar trend
to that found in other parts of the world (David Tracer, Personal communication).
Because parity is greater in Andean women, this can partially explain the effects of
ethnicity on birth weight, but these effects remain when parity is controlled for.

While not as strong as gestational age, the effect of parity is significant in all models
Figure 4.4Relationship between parity and birth weight
1 2 3 4 5 6 7 8
# Live births
This figure represents the relationship of birth weight with the number of live births
in our sample. Parity effects provide a significant (p<.001) contribution to birth
weight variance.

Maternal weight gain
Maternal weight gain can influence birth weight through the simple fact that
mothers who are able to gain more weight tend to have more resources to give to the
growing fetus. Lacking nutritional data, maternal weight gain provides the best
estimate of nutritional status. A major limitation exists in this measurement,
however, because maternal weight gain is based upon the difference between her
weight at first prenatal visit and her weight at delivery, and there are differences in
the week of first prenatal visit within our study population. Specifically, foreign
women have earlier initial visits. In our sample, absolute and percentage maternal
weight gain does not differ between the different ethnic categories when controlling
for the length of time between initial prenatal and final visits (Table 4.3). This is
potentially an important finding for our conclusions, because it may show that while
birth weights are lower among foreign bom mothers, that effect is likely not the
result of poor maternal nutrition.
Socioeconomic status
As a proxy for socioeconomic status, we used the nature of the mothers
employment, classified as skilled or unskilled. Women doing unskilled jobs
largely manual laborwere considered of a lower socioeconomic class than those
having skilled jobsprofessional or white collar positions. Compared with
unskilled jobs, skilled jobs often have higher pay and require less energy

expenditure. Table 4.3 shows the percentage of unskilled workers by ethnic group.
Socioeconomic status varies with ethnicity, with more Andeans having unskilled
jobs than foreign individuals, who more often have skilled office jobs.
Studies of the effect of socioeconomic status on birth weight at low altitude
consistently find lower birth weights among individuals in lower socioeconomic
strata (Parker et al 1994). Therefore, any effect of socioeconomic status in our
population would likely affect the results in the opposite direction of our gradients
(raising foreign birth weight and lowering Andean birth weight). Illustrating this,
when occupational class is controlled for, our findings are slightly more robust.
Table 4.3 Maternal weight gain, adjusted for week of first visit, and socioeconomic
Maternal ethnicity AA AM MM MF FF
Abs. weight gain 1.1+3 1.9+3 8.2+.3 7.5.5 7.8.6 p=.58
Percent weight 13.4+.6 13.8+.5 14.2+.3 12.6.9 13.1+1 p=.36
gain % unskilled 76 [70,82] 74 [68.1,79.9] 33 [30,36] 31 [22,40] 31 [21,41] p<.001

These results show that ethnic admixture, especially that transmitted maternally, has
an effect on birth weight that is independent of gestational age, parity, or maternal
age. These results are consistent with our hypothesis of differential effects of
maternal and paternal ethnicity upon birth weight, as well as ethnicity differences in
high altitude adaptation to pregnancy and matemal/fetal resource allocation. When
controlling for gestational age and parity, the mothers ethnicity effect upon birth
weight is significantly greater and more consistently linear, such that each added
dose of foreign admixture leads to a greater progressive decline in birth weight
than does the effect of paternal ethnicity.
While the fathers effect is not significantly linear, it is important to note that the
effect is marginally significant (p=.056) in an ANCOVA model. Further, the
paternal effect is marginally significant (p=. 091) when considering only Mestizo
mothers. Although paternal and maternal ethnicity are highly correlated in the full
data set, this marginal significance for the fathers effect in the Mestizo group may
demonstrate the existence of an effect of paternal ethnicity independent of maternal
ethnicity. This is important when considering that the fetus obtains half of its
genome from the father; this component could affect the fetuss ability to procure
resources from the mother. The lack of significance of the paternal effect in the
MANCOVA likely was affected by the noncontinuous distribution of cells when
both lineages are entered as full factors, and is thus difficult to interpret. The
marginal significance of the paternal effect (p=.090) in a categorical regression
model, which potentially reduces some of the error caused by that distribution,
further supports the potential presence of an independent paternal effect.

This thesis utilized data from a population-level study on birth weight at high
altitude to ascertain the relationship between ethnic admixture, parental lineage, and
birth weight. In doing so, the analyses presented highlight the importance of an
evolutionary and population-level perspective on human variation and adaptation.
The results demonstrate an independent effect of ethnic admixture upon birth weight
at high altitude. By breaking up ethnic admixture into more specific population units
than in previous studies, we were able to show that the relationship of birth weight
to ethnic admixture is, in fact, dependent upon the parents specific dose of
Andean and/or Foreign ethnicity. Further, these data show that there are differential
effects for maternal and paternal admixture upon birth weight at high altitude. The
maternal effect is consistently greater and presents a stronger linear trend than the
paternal effect across ethnic categories. Finally, an independent effect for paternal
ethnic admixture exists in the Mestizo group.
Ethnicity can reflect genetic differences within populations, as well as cultural ones.
Supported by previous work in Bolivia, we have used ethnic admixture as an
estimation of genetic admixture. The approximation of genetic admixture through
parental surnames potentially contributes to an understanding of the effect of natural

selection on adaptation to high altitude pregnancy (Chakraborty 1989). The effect of
ethnic admixture on birth weight, thus, likely indicates that genetic admixture
influences the altitude-associated reduction in birth weight.
This thesis has framed the interactions between ethnic admixture, parental lineage,
and birth weight in relation to the competing metabolic and genetic interests of the
mother, the fetus, and the father. If we postulate that individuals of foreign ancestry
are less well-adapted to high altitude (given their lower birth weights), and that this
effect is to some degree genetically-endowed, then it appears from our results that
attributes of the maternal genome affecting the uterine environment dictate fetal
growth to a greater degree than do paternal factors. This could occur because of the
genetic effects of adaptation to high altitude affecting the physiology of the
mothers uterine environment, but it could also involve a conflict of maternal and
paternal growth factors that operate through genomic imprinting. In either case, the
greater importance of the maternal lineage is consistent with life history
considerations for pregnancy. Because mothers at high altitude are already under
hypoxic stress, a greater vasculature and a correspondingly greater uterine artery
blood flow and nutrient transfer to the fetus may be required than at sea level.
Recent migrants to high altitude may therefore have less metabolic resources to give
to the growing offspring, and thus demonstrate poorer physiological adaptation to
In contrast, paternal ethnicity is not as strong or consistently linear across ethnic
categories as maternal ethnicity. When considering only Mestizo individuals,
paternal ethnic admixture does display a marginally significant effect upon birth
weight, but the pattern for the mothers effect to be stronger than the fathers
remains. Still, the paternal effect is potentially important in that birth weights
classified paternally are greater than those classified maternally across the majority

of ethnic categories. This paternal effect is consistent with David Haigs genetic
conflict theory of pregnancy and may suggest the operation of paternal genetic
factors in the fetus. As noted above, because pregnant women at high altitude are
energetically more stressed than those at sea level, scarce metabolic resources will
be heavily contested. Paternal parent-of-origin effects likely dictate some of the
behavior of cells in the maternal-fetal interface affecting fetal growth (e.g., IGFII) in
order to capture greater resources. However, lacking direct genetic evidence, it
appears somewhat difficult to accurately gauge the operation of paternal genes on
fetal growth between ethnic categories.
Without direct genetic evidence, we need to be cautious in our assignment of
genetic differences between ethnicity groups and parental lineages. A complete
determination of parent-of-origin effects at high altitude requires a full pedigree
analysis, including complete family, parental and sib phenotypic data. Aggregated
population-level data can elucidate trends through broad comparisons; however, the
ability to witness genomic imprinting at the population level is elusive. These data
do indicate that maternal ethnicity, which includes her genetic endowment as well
as her uterine physiology and culturally-influenced characteristics, is a better
predictor of offspring birth weight than the paternal ethnic lineage. Unfortunately,
these data cannot tell us if this is a result of a genetic conflict between mother and
fetus, if it is due to differential adaptation to the maternal environmental under
conditions of hypoxia, or if there is some interaction between these effects. Further,
in the absence of heritability studies, it remains difficult to ascertain the amount of
variation due to genetic factors in this population.
There exist limitations in the use of parental surnames for determining admixture, as
well. By separating doses of ethnicity, we think we have been able to obtain a
fairly accurate proxy measure for genetic admixture. However, we must recognize

that our surname data are a snapshot in time and an inaccurate one at that, because
maternal surnames are lost in each generation, and thus only paternal surnames are
retained after two generations. Further, surnames can be changed arbitrarily and
thus their utility in determining genetic relatedness lost.
A further limitation exists in that individuals with greater resources may be more
likely to descend to give birth, and also have access to resources (e.g. specialized
medical services) that permit some babies to survive that would otherwise have
died. This survivor effect could potentially remove small birth weight babies from
the population because those individuals with adequate resources left La Paz to give
birth at a lower altitude. Cesarian sections operate as a potential confounder in the
opposite direction, in that cesarian-section babies will artificially weigh a bit less
than if they went to term. Because they tend to have greater financial resources,, a
greater percentage of foreign individuals have cesarian sections. This factor likely
accounts for a portion of the ethnicity effect on birth weight, but when gestational
age is controlled for that effect remains.
Even considering these limitations, our results robustly suggest a greater effect for
maternal than paternal ethnicity on birth weight at high altitude. In controlling for
gestational age and parity, we removed two of the primary factors also affecting
birth weight in this population. It is important to note that from both the life-history
and maternal-offspring conflict perspectives, the length of gestation will likely be
influenced by maternal metabolic supply and fetal demand. Given the metabolic
crossover hypothesis (i.e., the notion that parturition will occur when fetal needs
outstrip the maternal ability to meet them), the effects of ethnicity at high altitude
Upon gestational age are similar in direction to their effects upon birth weight.

Social and behavioral variables that could affect the influence of ethnicity on birth
weight include mating practices, diet, and socioeconomic status. Our data display
significant assortative mating based on ethnicity, which limits the more discordant
forms of ethnic admixture. Diet and nutritional status are likely not implicated in the
effects we have observed, because maternal weight gain does not differ significantly
across ethnic categories. Finally, low socioeconomic status, which has been shown'
at low altitude to be related to low birth weight, is more prevalent among Andeans
and thus this factor cannot explain their heavier birth weights. After appropriately
addressing these potential confounders, we can assert that our observed effects of
ethnicity upon birth weight are largely independent of these factors.
In sum, genetic differences between ethnic ancestry groups remain as a likely
important, independent factor affecting birth weight at high altitude. At the
population level, it is possible that genetic factors have been selected for in
populations that have resided in the high altitude environment for a longer period of
time, which, in turn, permit normal fetal growth and improved chances for
reproductive success at high altitude. Evolutionary factors also affect resource
allocation at the individual level. This study has considered individual factors by
framing the competing evolutionary drives of mothers and fathers through the
perspective of life history theory, which dictates optimal resource allocation.
Genomic imprinting, postulated to result from intergenomic conflicts over that
resource allocation.
Pregnancy presents a critical period that likely guides the operation of natural
selection. This thesis has reviewed how interactions of biological and cultural
mechanisms affect adaptation to pregnancy at high altitude and has presented novel
analyses to understand the transmission of adaptation between generations.
Adaptation to pregnancy at high altitude appears to be closely related to the amount

of genetic admixture received from individuals of longer-resident populations, as
well as to the specific parental lineage through which that admixture is transmitted.
A further contribution of this thesis has been to illustrate how genomic imprinting
may be a factor influencing genetic adaptation to pregnancy at high altitude. To do
this, we considered the paternal effect of admixture on birth weight. In the full
sample, birth weights classified paternally are greater than those classified
maternally for all ethnic categories except for the Mestizo group. The difference in
birth weight between paternal and maternal classifications, while slight, is consistent
with the direction of growth effects predicted by David Haigs conflict theory. In
other words, paternally-transmitted alleles may be operating to increase birth
weight. Perhaps more importantly, in the Mestizo group, paternal ethnicity
independently influences birth weight. This may also indicate an effect of
paternally-transmitted alleles upon birth weight.
The effects of genomic imprinting have not previously been demonstrated at the
population level in humans, and here they are considered cautiously. Genomic
imprinting likely operates as an intergenerational mechanism for allocating scarce
resources between mother and fetus. Genomic imprinting provides a complex, non-
Mendelian perspective on an evolutionary understanding of gene-environment
interactions. Phenotypes can be viewed as the result not just of the developmental
interaction between genomes and the environment, but also of intergenerational and
inter-gender genomic conflicts that mediate the timing and allocation of resources to
various growth processes. In our study, birth weight at high altitude is affected by
the relative proportion of admixture from long-term high altitude populations, as
well as the parent from which the admixture is received.

Especially in a resource-scarce environment such as high altitude, an individual
organisms ability to procure resources for itself must be viewed as balanced against
the evolutionary trajectories and fitness concerns of those with whom it must share
limited resources. As Haig states, evolutionary models of optimal reproductive
investment predict that the birth weight that maximizes a mothers total number of
surviving offspring may be less than the weight that maximizes an individual babys
chance of survival (Haig 2001:492). Parent-offspring conflict theory predicts that
any given offspring will aim to maximize parental investment, but this is ultimately
determined by selective factors in both the mothers and fathers genetic ancestry.
This study further emphasizes the importance of an evolutionary approach to
determinants of health. Especially in resource scarce environments like high
altitude, both mother and fetus struggle to meet their metabolic needs. Paternal
factors acting in fetal cells, further, operate to procure resources in opposition to
maternal constraints. These paternal effects can be viewed as protective of the fetus;
in our study, this is consistent with heavier birth weights when classified paternally.
In considering evolutionary history, these effects can be framed as intergenerational
competitions for scarce resources. This interplay between evolutionary biology and
health determinants is important in understanding how adaptations to resource
scarcity manifest as health problems. Further, the idea of resource conflict is
important to keep in mind, because what is considered adaptive by evolutionary
logic will not always be adaptive for the individual. As Chisholm states, the lesson
from life history theory is that people, like other organisms, are not evolved to
maximize health, vigor, or lifespan, but ultimately, to have descendants (Chisholm
The view of individuals as subject to evolutionary forces, featuring the conflicting
strategies of genes, frightens many holistic thinkers who would prefer to view

individuals as being more autonomous. This leads some to resist biology and the
unpopularity of a genes eye view (Terleph 2000). However, a reductionistic
viewpoint is important for modem medicine; much of our current clinical conditions
involve combating the legacy of our own evolutionary history as written in our
genes. Importantly, though, the incorporation of social and cultural factors and their
interactionsby structuring mating in populations, the demographic influences on
reproduction, and the differential availability of resources including access to health
carewith life history strategies provides a more textured understanding of
intergenerational resource allocation. This study follows those that would support a
public health policy more informed by life history theory and a recognition of the
trade-offs inherent in these interactions (Chisholm 1995; Tracer 2002).

Atkin, Karl
2003 Ethnicity and the politics of the new genetics: principles and
engagement. Ethnicity and Health 8(2):91-109
Barker, DJP
1999 Fetal programming and public health. In Fetal Programming:
Influences on Development and Disease in Later Life. OBrien, PMS,
Wheeler T and Barker DJP, ed. pp. 3-11. London: RCOG Press
Beall, Cynthia
2000 Tibetan and Andean patterns of adaptation to high-altitude hypoxia.
Human Biology 72(l):201-228
Borgerhoff Mulder, Monique
1992 Reproductive decisions. In Smith, Eric and Bruce Winterhalder, eds.
Evolutionary Ecology and Human Behavior. New York: Aldine de Gruyter
Cartmill, Matt
1999 The status of the race concept in physical anthropology. American
Anthropologist 100(3):651-660
Chakraborty, Ranajit, Sara A. Barton, Robert E. Ferrell, and William J. Schull
1989 Ethnicity determination by names among the Aymara of Chile and
Bolivia. Human Biology 61 (2): 159-177

Chamock-Jones, D, P. Kaufman and T Mayhew
2004 Aspects of human fetoplacental vasculogenesis and angiogenesis.
Placenta 25:103-113
Chamov, Eric
1993 Life History Invariants. Oxford: Oxford University Press
Chisholm, James
1993 Death, hope, and sex: life history theory and the development of
reproductive strategies. Current Anthropology 34:1-24
Chisholm, James
1995 Life history theory and life style choice: implications for Darwinian
medicine. Perspectives in Human Biology 1:19-26
Coall, David and James Chisholm
2003 Evolutionary perspectives on pregnancy: maternal age at menarche
and infant birth weight. Social Science and Medicine 57:1771-1781
Conley, Dalton and Niel Bennett
2000 Race and the inheritance of low birth weight. Social Biology 47:1-2
Ellison, Peter
1990 Human ovarian function and reproductive ecology: new hypotheses.
American Anthropologist 92:933-952
Ellison, Peter
2001 On fertile ground. Cambridge: Harvard University Press.

Ellison, Peter
2003 Energetics and reproductive effort. American Journal of Human
Biology 15(3):342-351
Georgiades, Pantelis, Maria Watkins, Graham Burton, and Anne Ferguson-Smith
2001 Roles for genomic imprinting and the zygotic genome in placental
development. PNAS 98:4522-4527
Gangestad, Steven, Jeffrey Simpson
2000 The evolution of human mating: trade-offs and strategic pluralism.
Behavioral and Brain Sciences 23:573-644
Goos, Lisa, Irwin Silverman
2001 The influence of genomic imprinting on brain development and
behavior. Evolution and Human Behavior. 22:385-407
Gonzales, GF, Guerra-Garcia, R.
1993 Caracteristicas hormonales y antropometricas del embarazo y del
recien nacido en la altura. In Gonzales, GF, ed. Reproduccion Humana en la
Altura. Lima, Peru: Consejo Nacional de Ciencia y Tecnologia.
Haig, David
1993 Genetic conflicts in human pregnancy. The Quarterly Review of
Biology. 68:495-519

Haig, David
1996 Gestational drive and the green-bearded placenta. Proceedings of the
National Academy of Science 93:6547-6551
Haig, David
1999 Genetic conflicts of pregnancy and childhood In Steams, Stephen,
ed. Evolution in Health and Disease. New York: Oxford University Press
Haig, David
2000 The kinship theory of genomic imprinting. Annual Review of
Ecology and Systematics 31:9-32
Haig, David
2002 Genomic imprinting and kinship. New Jersey: Rutgers University
Haig, David
2003 Meditations on birth weight: is it better to reduce the variance or
increase the mean? Epidemiology 14:490-492
Hiby, Susan, Maria Lough, E. Barry Keveme, M. Azim Surani, Yung Wai Loke,
and Ashley King
2001 Paternal monoallelic expression of PEG3 in the human placenta.
Human Molecular Genetics 10:1093-1100
Hill, Kim and Hillard Kaplan
1999 Life history traits in humans: theory and empirical studies. Annual
Review of Anthropology 28:397-430

Hurst, Laurence, Gilean McVean
1997 Growth effects of uniparental disomies and the conflict theory of
genomic imprinting. Trends in Genetics 13:436-443
Iwasa, Yoshi
1998 The conflict theory of genomic imprinting: how much can be
explained? Current Topics in Developmental Biology 40:255-293
Jensen, Gwenn and Loma Moore
1997 The effect of high altitude and other risk factors on
birthweight:independent of interactive effects? American Journal of Public
Health 87:1003-1007
Kalmijn, Matthijs
1994 Assortative mating by cultural and economic occupational status.
The American Journal of Sociology 100:422-452
Keyes, Linda E., Fernando Armaza, Susan Niermeyer, Enrique Vargas, David A.
Young, and Loma G. Moore
2003 Intrauterine growth restriction, preeclampsia, and intrauterine
mortality at.high altitude in Bolivia. Pediatric Research 54(l):20-25
Kline J, Stein Z, and Susser, M.
1989 Conception to birth: epidemiology of prenatal development. New
York: Oxford

Kramer, M.S.
1987 Determinants of low birth weight: methodological assessment and
meta-analysis. Bulletin of the World Health Organization 65:663-737
Kono, Tomohiro, Yayoi Obata, Quiong Wu, Katsutoshi Niwa, Yukiko Ono, Yuji
Yamamoto, Eun Sung Park, Jeong-Sun Seo, and Hidehiko Ogawa
2004 Birth of parthenogenetic mice that can develop to adulthood. Nature
LeCroy, Dori and Peter Miller
2000 Evolutionary perspectives on human reproductive behavior. Annals
of the New York Academy of Sciences 907:1-233
Lewontin, R.C.
1972 The apportionment of human diversity. In Evolutionary Biology,
Vol. 6, Dobzhansky, T, M.K. Hecht, and W.C. Steers, eds. New York:
Lindsay, Robert, Sayuko Kobes, William Knowler, Robert Hanson
2002 Genome-wide linkage analysis assessing parent-of-origin effects in
the inheritance of birth weight. Human Genetics 110:503-509
Magnus, P, K. Berg, T. Bjerkdal and W. Nance
1984 Parental determinants of birth weight. Clinical Genetics 26:397-405
Magnus, P, H.K. Gjessing, A. Skrondal, R. Skjaerven
2001 Paternal contribution to birth weight. Journal of Epidemiology and
Community Health 55:873-878

Mawson, Jacinda T., Barry Braun, Paul R. Rock, Loma G. Moore, Robert Mazzeo,
and Gail E. Butterfield
2000 Women at altitude: energy requirement at 4,300 m. Journal of
Applied Physiology 88:272-281
Molnar, Stephen
1998 Human Variation. New Jersey: Prentice Hall
Moore, Loma
2001 Human genetic adaptation to high altitude. High Altitude Medicine
and Biology 2:257-279
Moore, Gudmn, Sayeda Abu-Amero, Gill Bell, Emma Wakeling, Amanda
Kingsnorth, Philip Stanier, Eric Jauniaux, and Simon Bennett
2001 Evidence that insulin is imprinted in the human yolk sac. Diabetes
Mochizuki, Atsushi, Yasuhiko Tekeda, Yoh Iwasa
1996 The evolution of genomic imprinting. Genetics 144:1283-1295
Moore, Loma, Susan Niermeyer, Stacy Zamudio
1998 Human adaptation to high altitude: regional and life-cycle
perspectives. Yearbook of Physical Anthropology 41:25-64
Morison, Ian and Anthony Reeve
1998 A catalogue of imprinted genes and parent-or-origin effects in
humans and animals. Human Molecular Genetics 7:1599-1609

Ohlsson, Richard
1999 Genomic Imprinting: An Interdisciplinary Approach. Springer:
Pan American Health Organization (PAHO)
2000 Health Analysis: Country Health Profiles. Electronic Document.
Parker J.D., K.C. Schoendorf, and J.L. Kiely
1994 Associations between measures of socioeconomic status and low
birth weight, small for gestational age, and premature delivery. Ann.
Epidemiology 4(4):271-8
Pike, Ivy
2001 The evolutionary and ecological context of human pregnancy. In
Ellison, Peter, ed. Reproductive Ecology and Human Evolution. New York
Aldine de Gruyter
Price, Kay and C.L. Coe
2000 Maternal constraint on fetal growth patterns in the rhesus monkey:
the intergenerational link between mothers and daughters. Human
reproduction 15:452-464
Ptak, Susan E. and Molly Przeworski
2002 Evidence for population growth in humans is confounded by fine-
scaled population structure. Trends in Genetics 18(11):559-563

Reik, Wolf and Azim Surani
1997 Genomic Imprinting. New York: Oxford University Press
Reik, Wolf and Jom Walter
2001 Genomic imprinting: parental influence on the genome. Nature
Reviews 2:21-32
Reik, Wolf, Miguel Constancia, Abigail Fowden, Neil Anderson, Wendy Dean,
Anne Ferguson-Smith, Benjamin Tycko and Colin Sibley
2003 Regulation of supply and demand for maternal nutrients in mammals
by imprinted genes. Journal of Physiology 547:35-44
Rockwell LC, Vargas, E and Loma Moore
2003 Human physiological adaptation to pregnancy. American Journal of
Human Biology 15(3):330-41
Sanchez-Andres, A. and M.S. Mesa
1994 Assortative mating in a Spanish population: effects of social factors
and cohabitation time. Journal of Biosocial Science 26:441-450
Spencer, Hamish, Marcus Feldman, and Andrew Clark
1998 Genetic conflicts, multiple paternity, and the evolution of genomic
imprinting. Genetics 148:893-904
Spencer, Hamish, Andrew Clark, and Marcus Feldman
1999 Genetic conflicts and the evolutionary origin of genomic imprinting.
TREE 14:197-201

Spencer, Hamish
2000 Population genetics and evolution of genomic imprinting. Annual
review of Genetics Annual :457
Templeton, Alan
1999 Human races: a genetic and evolutionary perspective. American
Anthropologist 100(3):632-650
Terleph, Thomas
2001 Resisting biology. Annals of the New York Academy of Sciences
Tracer, David
2002 Somatic versus reproductive energy allocation in Papua New Guinea:
life history theory and public health policy. American Journal of Human
Biology 14:621-626
Trevathan, Wenda, E.O. Smith, James McKenna
1999 Evolutionary Medicine. New York: Oxford
Trivers, Robert
2002 Natural Selection and Social Theory. New York: Oxford
Weisstein, Anton and Hamish Spencer
2003 The evolution of genomic imprinting via variance minimization: an
evolutionary genetic model. Genetics 165:205-222

Wood, James
1994 Dynamics of Human Reproduction. New York: Aldine de Gruyter