Received: 14 January 2023 Revised: 24 May 2023 Accepted: 8 June 2023
DOI: 10.1002/ajhb.23943
OR I G I N A L AR T I C L E
Human milk immune factors, maternal nutritional status,
and infant sex: The INSPIRE study
Beatrice Caffé1 | Aaron Blackwell1 | Bethaney D. Fehrenkamp2,3 |
Janet E. Williams4 | Ryan M. Pace2 | Kimberly A. Lackey2 | Lorena Ruiz5,6 |
Juan M. Rodríguez7 | Mark A. McGuire4 | James A. Foster8 |
Daniel W. Sellen9 | Elizabeth W. Kamau-Mbuthia10 | Egidioh W. Kamundia10 |
Samwel Mbugua10 | Sophie E. Moore11,12 | Andrew M. Prentice12 |
Linda J. Kvist13 | Gloria E. Otoo14 | Rossina G. Pareja15 | Lars Bode16,17 |
Dubale Gebeyehu18 | Debela K. Gindola18 | Sarah Boothman19 |
Katherine Flores1 | Michelle K. McGuire2 | Courtney L. Meehan1
1Department of Anthropology, Washington State University, Pullman, Washington, USA
2Margaret Ritchie School of Family and Consumer Sciences, University of Idaho, Moscow, Idaho, USA
3Washington, Wyoming, Alaska, Montana, Idaho (WWAMI) Medical Education Program, University of Idaho, Moscow, Idaho, USA
4Department of Animal, Veterinary and Food Sciences, University of Idaho, Moscow, Idaho, USA
5Department of Microbiology and Biochemistry of Dairy Products, Instituto de Productos L
acteos de Asturias (IPLA-CSIC), Villaviciosa, Spain
6Instituto de Investigacio
n Sanitaria del Principado de Asturias (ISPA), Microhealth Group, Oviedo, Spain
7Department of Nutrition and Food Science, Complutense University of Madrid, Madrid, Spain
8Department of Biological Sciences, University of Idaho, Moscow, Idaho, USA
9Department of Anthropology, University of Toronto, Toronto, Ontario, Canada
10Department of Human Nutrition, Egerton University, Nakuru, Kenya
11Department of Women and Children's Health, King's College London, London, UK
12MRC Unit The Gambia at the London School of Hygiene and Tropical Medicine, Fajara, Gambia
13Faculty of Medicine, Lund University, Lund, Sweden
14Department of Nutrition and Food Science, University of Ghana, Accra, Ghana
15Nutrition Research Institute, Lima, Peru
16Larsson-Rosenquist Foundation Mother-Milk-Infant Center of Research Excellence, University of California, San Diego, La Jolla, California, USA
17Department of Pediatrics, University of California, San Diego, La Jolla, California, USA
18Department of Anthropology, Hawassa University, Hawassa, Ethiopia
19School of Biological Sciences, Washington State University, Pullman, Washington, USA
Correspondence
Courtney L. Meehan, Department of Abstract
Anthropology, Washington State Objectives: Breastfeeding is an energetically costly and intense form of human
University, Pullman, Washington, USA. parental investment, providing sole-source nutrition in early infancy and bio-
Email: cmeehan@wsu.edu
active components, including immune factors. Given the energetic cost of
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
© 2023 The Authors. American Journal of Human Biology published by Wiley Periodicals LLC.
Am J Hum Biol. 2023;e23943. wileyonlinelibrary.com/journal/ajhb 1 of 14
https://doi.org/10.1002/ajhb.23943
2 of 14 CAFFE
 ET AL.
Funding information
European Comission; Health Equity lactation, milk factors may be subject to tradeoffs, and variation in concentra-
Research Center, Washington State tions have been explored utilizing the Trivers-Willard hypothesis. As human
University; Medela; National Institutes of milk immune factors are critical to developing immune system and protect
Health; National Science Foundation; U.S.
Department of Agriculture infants against pathogens, we tested whether concentrations of milk immune
factors (IgA, IgM, IgG, EGF, TGFβ2, and IL-10) vary in response to infant sex
and maternal condition (proxied by maternal diet diversity [DD] and body
mass index [BMI]) as posited in the Trivers-Willard hypothesis and consider
the application of the hypothesis to milk composition.
Methods: We analyzed concentrations of immune factors in 358 milk samples
collected from women residing in 10 international sites using linear mixed-
effects models to test for an interaction between maternal condition, including
population as a random effect and infant age and maternal age as fixed effects.
Results: IgG concentrations were significantly lower in milk produced by
women consuming diets with low diversity with male infants than those with
female infants. No other significant associations were identified.
Conclusions: IgG concentrations were related to infant sex and maternal diet
diversity, providing minimal support for the hypothesis. Given the lack of asso-
ciations across other select immune factors, results suggest that the Trivers-
Willard hypothesis may not be broadly applied to human milk immune factors
as a measure of maternal investment, which are likely buffered against pertur-
bations in maternal condition.
1 | INTRODUCTION Reflecting local ecology (Klein et al., 2018), human
milk is thus uniquely suited to prime newborns for their
The composition and abundance of human milk immune environment (Andreas et al., 2015; Brandtzaeg, 2003). In
factors are known to be reflective of a mother's environ- the early postpartum period, a time when infants'
mental conditions, including her physical environment immune systems remain naive (Ballard & Morrow, 2013),
(Peroni et al., 2010), geographical location (Amoudruz mothers transfer immunity to their infants via the provi-
et al., 2009; Holmlund et al., 2010; Munblit et al., 2016; Ruiz sion of colostrum and milk (Andreas et al., 2015;
et al., 2017), subsistence strategy (Klein et al., 2018), and Brandtzaeg, 2003). Lactation, however, is an energetically
microbial exposure (McGuire et al., 2021; Prentice, costly form of parental investment (600 kcals per day;
Watkinson, et al., 1984; Tomici
c et al., 2010). Human milk Butte & King, 2005; Lunn, 1994) which involves diverting
immune factors are also associated with a variety of mater- available energy away from maternal needs to the synthe-
nal and infant characteristics including delivery mode sis of milk for the offspring, an investment that can span
(Groer et al., 2004; Striker et al., 2004), gestational age several months to years (Valeggia & Ellison, 2001).There-
(Castellote et al., 2011), mastitis (Castro et al., 2022; fore, lactation and the composition of human milk are
Espinosa-Martos et al., 2016; Hassiotou et al., 2013; Hunt likely subject to energetic tradeoffs and may be respon-
et al., 2013; Prentice et al., 1985), time postpartum sive to proximate cues in the environment and from the
(Ballard & Morrow, 2013; Czosnykowska-Łukacka mother-infant dyad.
et al., 2020; Goldman et al., 1983; Moles et al., 2015;
Munblit et al., 2016; Prentice, Prentice, et al., 1983; Turin &
Ochoa, 2014), pasteurization method (Escuder-Vieco 1.1 | Evolutionary approaches to human
et al., 2018; Espinosa-Martos et al., 2013), and infant illness parental investment and human milk
(Breakey et al., 2015; Bryan et al., 2007; Hassiotou composition
et al., 2013; Riskin et al., 2012). Maternal diet (Böttcher
et al., 2008; Urwin et al., 2012) and nutritional status (Lu Evolutionary approaches examining how parental invest-
et al., 2018; Miranda et al., 1983) are also frequently associ- ment may interact and influence human milk composi-
ated with variation of immune factor concentrations tion (e.g., Fujita et al., 2019; Powe et al., 2010) have
in milk. typically centered on whether there are associations
CAFFE
 ET AL. 3 of 14
between a mother's milk composition and the sex of her mothers with daughters. Conversely, mothers with low
infant (e.g., do male infants receive milk that has higher MUAC and with daughters had higher concentrations of
energy density than female infants?). A subset of these sIgA in their milk than did mothers with sons (Fujita
papers has directly or indirectly examined the Trivers- et al., 2019). However, among Filipino mothers, Quinn
Willard hypothesis or the interaction effects between (2013) found no evidence of an infant sex bias in either
maternal condition and infant sex on milk composition fat, lactose, protein, or energy content of the milk of
(Eckart et al., 2021; Fujita et al., 2012, 2019; Powe mothers when accounting for maternal BMI. In general,
et al., 2010; Quinn, 2013). Trivers-Willard posits that studies testing the Trivers-Willard hypothesis in post-
mothers attain greater reproductive success through birth parental investment contexts have been challenged
investing in offspring that are most likely to obtain by a lack of cross-cultural data, small sample sizes, and
greater fitness (Trivers & Willard, 1973). Briefly, the the operationalization of parental investment and mater-
hypothesis suggests that maternal condition will influ- nal condition (Cronk, 2007; Thouzeau et al., 2022). Nev-
ence offspring condition and argues males born to ertheless, the hypothesis continues to be tested across a
mothers in poor condition (e.g., underweight, malnour- variety of parental investment domains (e.g., Lynch
ished) will experience poor condition and have more et al., 2018; Pink et al., 2017; Song., 2018).
challenges in obtaining mates and reproducing than Here, utilizing a previously collected dataset includ-
males born to mothers in good condition. Mothers in ing a range of human milk immune factors (Immuno-
poor condition are predicted to skew their investment globulin A, Immunoglobulin G, Immunoglobulin M,
towards female infants, who experience lower levels of Epidermal Growth Factor, Transforming Growth Factor
reproductive variability, while mothers in good condi- Beta 2, and Interleukin-10) and standardized maternal
tions are expected to skew investment towards male off- condition variables (maternal BMI and diet diversity
spring because, under those conditions, males may scores) from a multi-cohort, international human milk
achieve greater reproductive success than their female composition study (the INSPIRE study), we test the
counterparts (Trivers & Willard, 1973). TriversWillard hypothesis and consider its utility as an
Human milk is viewed as a form of parental invest- evolutionary framework in human milk studies. The
ment, whereas milk synthesis under certain circum- application of the Trivers-Willard hypothesis to human
stances can conflict with maternal interest. For example, milk immune factors would suggest that resource limited
the mobilization of body stores to source calcium for milk mothers may face a tradeoff between caloric investment
production leads to temporal bone mineral loss in in milk and production of milk immune factors. These
mothers until after weaning (Kalkwarf & Specker, 2002). might covary due to phenotypic correlation, but immune
Currently, it remains uncertain whether milk immune factors might be expected to be prioritized secondary to
factors are costly enough to produce to be considered basic caloric needs. It is important to note that costs to
maternal investment. However, several studies on IgA mothers of producing human milk immune factors are
suggest that it is a possibility. For instance, certain situa- unknown and therefore operationalizing human milk
tions such as extreme undernutrition (Miranda immune factors as investment is challenging. Neverthe-
et al., 1983), reduced food availability (Weaver less, this study represents the largest, methodologically
et al., 1998) and intensive exercise (Gregory et al., 1997) consistent test of this hypothesis to our knowledge.
have shown a decrease in milk IgA concentrations. Therefore, if the investment variation predictions of
Applications of the Trivers-Willard hypothesis to Trivers-Willard hypothesis apply to human milk immune
human milk composition, however, have yielded conflict- factors, we should expect concentrations of immune fac-
ing results. Among well-nourished women in the tors will be lower in milk produced by underweight or
United States, Powe et al. (2010) found the milk produced overweight/obese women or women consuming diets
by women breastfeeding male infants was higher in with lower diversity (indicating poor nutritional status,
energy density than the milk of mothers with female i.e., poor condition) with sons than those in milk pro-
infants. Fujita et al. (2012) found higher fat content in duced by mothers in poor conditions with daughters.
the milk of poor Kenyan mothers with female infants, Specifically, we test whether Immunoglobulin A (IgA),
while economically secure Kenyan mothers produced Immunoglobulin G (IgG), Immunoglobulin M (IgM),
milk with higher fat content for their sons. Results from Epidermal Growth Factor (EGF), Transforming Growth
another study in the same population showed that Factor Beta 2(TGFβ2), and Interleukin 10 (IL-10) concen-
mothers with high mid-upper arm circumference trations in the milk of mothers with infants varied
(MUAC) with sons had higher concentrations of secre- according to the interaction between infant sex and
tory Immunoglobulin A (sIgA) in their milk than did maternal condition.
4 of 14 CAFFE
 ET AL.
2 | METHODS cleaned twice with pre-packaged soap wipes by gloved
hands either by researchers or participants. At least
2.1 | Study populations 20 mL of milk were expressed either by hand or by use of
an electric pump (Medela, Inc. McHenry, IL) into a
Data used in this analysis were previously collected as single-use, sterile polypropylene milk container with a
part of a prospective multi-cohort, international, cross- polybutylene cap (Medela, Inc.) or polypropylene speci-
sectional study (the INSPIRE study) (Lackey et al., 2019; men containers with polyethylene caps (VWR Interna-
Lane et al., 2019; McGuire et al., 2017; McGuire et al., tional, LLC). Milk was placed on ice or in a cold box
2021; Pace et al., 2021; Ruiz et al., 2017). Data were until aliquoted (within 1 h of the collection). Aliquots of
collected on 410 breastfeeding women and their milk were frozen at 
20C and then shipped on dry ice
infants between May 2014 and April 2016. The original to Complutense University in Madrid, where immuno-
sample set included two European (Spanish and Swed- logical assays were performed.
ish), one South American (Peruvian), two North Amer-
ican (United States: Southeastern Washington/
northwestern Idaho and southern California), and six 2.3 | Immune factors and
sub-Saharan African (rural and urban Ethiopian, rural immunological assays
and urban Gambian, Ghanaian, and Kenyan) cohorts.
Detailed descriptions of these populations have been We have previously reported the detailed description of
published previously (see McGuire et al., 2017; Ruiz the immunological assay and the wide range of concen-
et al., 2017). The rural Ethiopian cohort (n = 40) was trations of immune factors in milk of participants in the
not included in this study due to variation in sample INSPIRE study (Ruiz et al., 2017). Briefly, samples from
preservation approaches that prevented immune factor all study sites were aliquoted and subjected to a single
analysis. freeze–thaw cycle and were analyzed by the same
Women did not have to be exclusively breastfeeding researchers using the same reagents and equipment to
to participate. Infants enrolled in the study were between reduce potential lab biases. To avoid defrosting cycles,
the ages of 2 weeks and 5 months. To be included in the individual aliquots were used for each analysis. Prior to
study, infants had to be reported healthy by their analyses, 1 mL samples were centrifuged at 800g for
mothers, defined as no reported illness (including fever, 15 min at 4C to remove the fatty layer and the interme-
vomiting, severe cough, rapid breathing, or diarrhea) diate aqueous phase was used for the immunoassay
7 days prior to enrollment, and have not been given anti- determinations as previously described (Espinosa-Martos
biotics in the month prior to participation. Mothers had et al., 2013). All assays were run in duplicate according to
to self-report as healthy, defined as having had no symp- manufacturer's instructions and standards curves were
toms of illness (including fever, vomiting, severe cough, performed for each analyte. Magnetic bead-based multi-
or diarrhea) in the 7 days prior to enrollment, not plex immunoassays were used to assess the concentra-
experiencing breast infection or atypical breast pain that tions of TGFβ2 and IL-10 (BioRad, Hercules, CA, USA).
they did not consider normal for lactation, and had not TGFβ2 was acid-activated prior to analysis, per manufac-
taken antibiotics in the 30 days before enrollment. turer recommendations. The RayBio Human EGF ELISA
Informed, verbal, or written consent (depending on kit was used to quantify EGF concentration (RayBiotech,
locale and the subject's literacy level) was acquired from Norcross, GA, USA). The concentrations of IgG, IgA, and
each participating woman. Ethics approvals were IgM were determined using the Bio-plex Pro Human Iso-
obtained for all procedures from each participating insti- typing Assay kit (BioRad, Hercules, CA, USA). Immuno-
tution and with overarching approval from the globulins (IgA, IgG, IgM) were measured in milligrams
Washington State University Institutional Review Board per liter, IL-10 in nanograms per liter, and EGF and
(IRB #13264). TGFβ2 in micrograms per liter (Ruiz et al., 2017). The
inter-assay coefficient of variation were below manufac-
turer's instructions for all immune markers and detection
2.2 | Milk collection limits of assays reported previously (Ruiz et al., 2017).
Immune factor concentrations were log transformed
All milk collection supplies (including gloves, soap wipes, prior to statistical analysis. IgA was undetectable for one
and collection containers) were standardized and pro- sample, TGFβ2 in one sample, and IL-10 in 118 samples.
vided to personnel at each site to help control for known These values were assigned a value of one half the lower
and unknown biases. The participant's study breast was limit of detection, after log transformation.
CAFFE
 ET AL. 5 of 14
2.4 | Maternal and infant EGF, TGFβ2, IL-10, maternal characteristics that serve as
characteristics, diet diversity scores and a proxy for maternal condition (maternal BMI, maternal
maternal BMI DDS), and infant sex (female or male). Each LME model
included population as a random effect and infant age
Maternal and infant age and infant sex were reported by (wk) and maternal age (y) as fixed effects. In the LME
participating mothers. Maternal BMI and diet diversity models, BMI and diet diversity scores were treated as
scores were calculated and used as a proxy for maternal continuous variables, with higher BMI or diet diversity
condition. BMI and diet were chosen because of previous indicating better condition. However, since very high and
studies that have tested Trivers-Willard (e.g., Fujita low BMIs may also indicate poor condition, we also
et al., 2012; Quinn, 2013) and because they represent the tested a second model with BMI as categorical variable
best available data in an existing data set. Diet diversity according to the classifications described above (under-
has been consistently associated with food insecurity weight, healthy and overweight/obese). Statistical signifi-
among breastfeeding and pregnant women (Kang cance was declared at P ≤ .05 and statistically trending at
et al., 2019; Na et al., 2016; Singh et al., 2020). Specifically P > .05–≤ .1. Additional models were run with the
in regard to maternal condition, DDS has been associated 240 datapoints from samples with IL-10 above the limit
with maternal thinness and anemia in rural Cambodia of detection; these models did not differ from the models
(McDonald et al., 2015). with the complete sample in terms of qualitative infer-
Maternal BMI was calculated from maternal height ences, so are not reported.
and weight measurements obtained by study personnel.
BMI classifications were made according to US Centers
for Disease Control and Prevention guidelines 3 | RESULTS
(CDC, 2021), with scores <18.5 considered underweight;
18.5 to <25 as healthy, 25 to <30 as overweight and ≥30 3.1 | Participant characteristics
as obese. We combined overweight and obese categories
with scores ≥25, defining the category as overweight/ Of the 410 mothers from which we obtained milk sam-
obese. ples in the original INSPIRE study, 358 mothers are
Diet diversity was characterized by creating a diet included in this study. As noted, data from rural Ethiopia
diversity score (DDS) for each subject based on the 2011 were excluded due to variation in sample preservation,
US Food and Agriculture (FAO) Guidelines for Measur- and an additional 12 individuals from other populations
ing Household and Individual Dietary Diversity were missing anthropometric (n = 11) or dietary
(Kennedy et al., 2011). Mothers were asked to recall their data (n = 1).
food consumption over the past 7 days against a checklist The mean maternal age was 27.8 years, ranging
of food groups. Data were used to create a DDS similar to between 18 and 46 years. Infant age ranged from 2.9 to
the women's diet diversity score included in the FAO 16.6 weeks with a mean of 9 weeks old. There were some
guidelines (Kennedy et al., 2011). For creation of the differences (e.g., BMI and maternal age) in participant
DDS, the 14 dietary categories from the original diet sur- characteristics among the cohorts; select characteristics
vey responses were aggregated into 8 consolidated catego- are summarized in Table S1. There were no significant
ries (i.e., starchy staples; vegetables; fruits; organ meat; differences in maternal age, maternal BMI, or maternal
meat and fish; eggs; legumes, nuts and seeds; milk and DDS between mothers of male and female infants
milk products). Responses were coded dichotomously, Table 1.
where for each category an affirmative response was
coded as one and a negative was coded as zero. The DDS
of each participant was calculated by summing the num- 3.2 | Maternal BMI, infant sex, and HM
ber of categories of consumed foods, with potential scores immune factor concentrations
ranging from 0 to 8.
Among underweight and overweight/obese women,
those with sons did not have lower concentrations of
2.5 | Statistical analysis immune factors in their milk compared to women in sim-
ilar condition with daughters. LME modeling indicated
All statistical analyses were performed using R (4.0.2). R no significant interactions between infant sex and mater-
package nlme (V 3.1-152) was used to create linear nal BMI (as a continuous variable) on concentrations of
mixed-effects (LME) models to examine the potential IgA, IgG IgM, EGF, or IL-10 (Table 2, Figure 1). How-
relationship between concentrations of IgA, IgG, IgM, ever, the interaction approached significance for TGFβ2
6 of 14 CAFFE
 ET AL.
TABLE 1 Maternal and infant characteristics by infant sex.
Total sample (n = 358) Female (n = 182) Male (n = 176) P value
Mothers
Maternal age, year 27.76 ± 6.19 27.34 ± 6.25 28.19 ± 6.12 0.193
Maternal BMI, kg/m2 24.60 ± 4.60 24.24 ± 4.24 24.98 ± 4.90 0.126
Maternal BMI category
Underweight 17.38 ± 0.90 17.33 ± 1.23 17.41 ± 0.57 0.163
Healthy 21.98 ± 1.73 24.24 ± 4.24 21.99 ± 1.81 0.616
Overweight/Obese 29.23 ± 3.67 28.67 ± 3.51 29.75 ± 3.76 0.476
Maternal DDS 6.11 ± 1.32 6.15 ± 1.31 6.06 ± 1.32 0.539
Infants
Infant agea, week 9.02 ± 2.89 9.15 ± 2.83 8.88 ± 2.93 0.375
Note: Values are given as mean ± standard deviation. P values were calculated from Welch's two sample t-test for continuous variables and Pearson's chi-
square for the categorical variables.
aInfant age is equivalent to maternal time postpartum.
TABLE 2 Linear mixed-effects models examining variation in human milk immune factor concentrations with maternal body mass
index (BMI) as a continuous predictor.
IgA (mg/L) IgG (mg/L) IgM (mg/L) EGF (μg/L) TGFβ2 (μg/L) IL-10 (ng/L)
Covariates β P β P β P β P β P Β P
(Intercept) 5.992 <0.001 4.963 <0.001 4.307 <0.001 1.940 <0.001 0.776 0.207 
0.312 0.783
Infant Sexa 
0.350 0.497 
0.306 0.374 
0.634 0.214 
0.270 0.272 
1.261 0.058 0.449 0.630
Maternal BMI 0.009 0.587 
0.016 0.151 
0.015 0.360 
0.004 0.603 
0.038 0.075 
0.005 0.872
Infant Ageb 
0.018 0.292 
0.003 0.776 
0.029 0.083 
0.010 0.232 0.048 0.029 0.011 0.728
Maternal Age 0.008 0.400 
0.014 0.015 
0.003 0.760 
0.004 0.408 
0.005 0.648 
0.009 0.581
Infant Sex: 0.010 0.631 0.010 0.471 0.025 0.229 0.011 0.268 0.048 0.070 
0.020 0.588
Maternal BMI
Random effects 0.551 0.875 0.673 0.585 0.690 0.866 0.370 0.418 0.355 1.129 2.316 1.582
Population
(Intercept,
residual)
Note: Values on the table are bolded to highlight signficant values.
aInfant male coded as 1, female coded as 0.
bInfant age is equivalent to maternal time postpartum.
(β = 0.048, P = .07, Table 2, Figure 1). Mothers with the use of BMI in linear terms. However, we present
BMIs <26.5 kg/m2 with female infants trended towards these results cautiously as this analysis may not be cross-
higher concentrations of TGFβ2 in their milk than culturally valid. Nevertheless, description of analysis and
mothers with male infants and BMIs <26.5 kg/m2 results are included in the supplementary (Table S2).
(Figure 1). Other immune factors in our linear models,
except for IL-10, followed the same overall pattern of
higher concentrations in milk of mothers of female 3.3 | Maternal diet diversity, infant sex,
infants with lower BMI relative to those with male and HM immune factor concentrations
infants, and vice versa for higher BMI, but the differences
were not significant. LME models showed an interaction between infant sex
Next, we ran models with BMI coded categorically to and maternal DDS on IgG concentration (β = 0.099,
investigate non-linear effects of underweight and over- P = .037; Table 3). Specifically, mothers consuming less
weight/obese BMIs. Results showed general support for diverse diets with sons had lower concentrations of IgG
CAFFE
 ET AL. 7 of 14
(A) (B) (C)
9 7
6
6
6
4
5
3
4
2
0 3
0
2
−3
20 30 40 20 30 40 20 30 40
BMI BMI BMI
(D) (E) (F)
6
4 4
2 2
4
0
0
2 −2
−2
−4
−4
0
20 30 40 20 30 40 20 30 40
BMI BMI BMI
FIGURE 1 Lines show predictions of linear mixed-effects models examining variation in human milk immune factor concentrations
with maternal body mass index (BMI) as a continuous predictor. Shaded areas are 95% confidence intervals for the effect. A. IgA (β = 0.010,
P = .631) B. IgG (β = 0.010, P = .471) C IgM (β = 0.025, P = .229) D. EGF (β = 0.011, P = .268) E.TGFβ2(β = 0.04, P = .268) F. IL-10
(β = 0.020, P = .588) All factors follow the same overall trend of higher concentrations in milk of mothers with male infants and higher BMI
scores except for IL-10, however no relationships are significant.
TABLE 3 Linear mixed-effects outputs for interaction between infant sex, immune factors and maternal diet diversity score (DDS).
IgA (mg/L) IgG (mg/L) IgM (mg/L) EGF (μg/L) TGFβ2 (μg/L) IL-10 (ng/L)
Covariates β P β P β P β P β P β P
(Intercept) 6.383 <0.001 4.923 <0.001 4.557 <0.001 1.642 <0.001 0.111 0.845 
1.460 0.186
Infant Sexa 
0.156 0.725 
0.671 0.024 
0.870 0.047 
0.037 0.861 
0.732 0.201 1.469 0.067
Maternal DDS 
0.032 0.575 
0.052 0.173 
0.096 0.084 0.035 0.197 
0.027 0.702 0.169 0.098
Infant Ageb 
0.021 0.222 
0.003 0.803 
0.031 0.062 
0.011 0.176 0.047 0.032 0.016 0.590
Maternal Age 0.009 0.282 
0.016 0.006 
0.003 0.727 
0.004 0.362 
0.008 0.462 
0.011 0.474
Infant sex: 0.009 0.902 0.099 0.037 0.137 0.051 0.007 0.845 0.107 0.243 
0.249 0.052
Maternal
DDS
Random 0.587 0.877 0.689 0.583 0.690 0.860 0.358 0.418 0.352 1.133 2.337 1.575
effects
Population
(Intercept,
residuals)
Note: Values on the table are bolded to highlight signficant values.
aInfant male coded as 1, female coded as 0.
bInfant age is equivalent to maternal time postpartum.
Log EGF (µg/L) Log IgA (mg/L)
Log TGFß2 (µg/L) Log IgG (mg/L)
Log IL−10 (ng/L) Log IgM (mg/L)
8 of 14 CAFFE
 ET AL.
(A) (B) (C)
9 7
6
6
6
4
5
3
4 2
0 3
0
−3 2
2 4 6 8 2 4 6 8 2 4 6 8
DDS DDS DDS
(D) (E) (F)
6 4 4
2
2
4
0
0
2 −2
−2
−4
0 −4
2 4 6 8 2 4 6 8 2 4 6 8
DDS DDS DDS
FIGURE 2 aDiet Diversity Score = Score depicting the relative diversity of food categories consumed by participants. Lines show
predictions of linear mixed-effects models examining interaction between infant sex, immune factors and maternal diet diversity score
(DDS). Shaded areas are 95% confidence intervals for the effect. A. IgA (β = 0.009, P = .902) B.IgG (β = 0.009, P = .037) C. IgM (β = 0.137,
P = .051) D.EGF (β = 0.007, P = .845) E. TGFβ2 (β = 0.107, P = .243) F. IL-10 (β = 
0.249 P = .052) All factors follow the same overall
trend of higher concentrations in milk of mothers with male infants and high diet diversity except for IL-10, however only IgG is significant.
in their milk compared to mothers in similar condition underweight or overweight/obese BMI or low DDS (poor
with daughters (Figure 2B). The interaction approached condition) with sons than in the milk of mothers with
significance in the same direction for IgM (β = 0.137, daughters and with underweight or overweight/obese
P = .051; Table 3). However, mothers with high DDS and BMI or low DDS.
male infants trended towards lower concentrations of IL- We found minimal support for the Trivers-Willard
10 in their milk compared to mothers with high DDS and hypothesis. Mothers consuming less diverse diets with
female infants (β = 
0.249, P = .052; Table 3, Figure 2F), male infants had significantly lower concentrations of
but it was not significant. IgG in their milk compared to mothers with less diverse
diets with female infants. This finding is consistent with
previous research that found female-biased investment
4 | DISCUSSION among mothers in poor condition (with varying measures
of condition, such as nourishment and economic stabil-
This study investigated maternal biological investment ity) in the form of energy content (Powe et al., 2010) and
via concentrations of human milk immune factors using fat content (Fujita et al., 2012) in humans. However, our
predictions from the Trivers-Willard hypothesis. We results did not find support for a Trivers-Willard effect
tested whether concentrations of select immune factors with any other immune factor tested with diet diversity
(IgA, IgG, IgM, EGF, TGFβ2, and IL-10) would vary or BMI as a measure of maternal condition. Additionally,
according to maternal condition and infant sex. Specifi- no associations were found to support the prediction that
cally, we examined whether concentrations of immune underweight and overweight/obese women would have
factors would be lower in the milk of mothers with lower concentrations of immune factors in their milk for
Log EGF (µg/L) Log IgA (mg/L)
Log TGFß2 (µg/L) Log IgG (mg/L)
Log IL−10 (ng/L) Log IgM (mg/L)
CAFFE
 ET AL. 9 of 14
their sons than women in similar conditions with order to conduct analyses cross-culturally and in multiple
daughters. sites around the world, we used a broad definition of
Researchers have proposed that immune factors in “healthy” which was determined by an inclusion/
milk are costly to produce, and the production of those exclusion criterion of no reported illness 7 days prior to
factors are subject to life-history tradeoffs or shifts in enrollment and no antibiotics use a month prior to par-
energy expenditure (Breakey et al., 2015; Miller, 2018). ticipation for mothers and infants and no symptoms of
The mechanism by which immune factors end up in breast infection or atypical breast pain for mothers. How-
human milk, however, may reflect their overall invest- ever, as seen in the data, variance existed in maternal
ment costs. Some immune factors (e.g., IgA) are synthe- condition (i.e., BMI and DDS) despite all women self-
sized in the mammary gland from B cells that migrate identifying as healthy. Additionally, DDS are a measure
from the gastrointestinal tract during pregnancy (Roux of nutritional diversity, but they do not provide quantita-
et al., 1977). Others (e.g., cytokines) are produced by tive intake or account for fats and oil, which are contribu-
mammary epithelial cells or by cells carried within the tors to energy stores (Kennedy et al., 2011) and therefore
milk, while some (e.g., IgG) are transferred from mater- may not be a comprehensive or definitive measure of
nal serum (Agarwal et al., 2011) and may be more costly nutritional status. As an indirect measure of body fat and
for mothers to produce. However, immune factors pro- obesity that relies solely on height and weight
duced locally in the mammary gland may be less costly (Rothman, 2008), BMI may be reflective of poor condi-
or require less energy for mothers to produce than those tion in terms of mother's overall health but may not
transferred from the maternal circulation (Butler reflect poor energy availability. Moreover, there are
et al., 2015). The exact costs of IgG, the only significant numerous known and potential unknown factors that
finding that supported the hypothesis, are unknown due influence variation in human milk immune factor con-
to the lack of research on the impact of IgG transferring centrations (e.g., potential unknown genetic influences).
from maternal circulation to other sites, including the Here, we only tested a limited range of those factors
placenta and mammary gland. Therefore, if variation in through measures that were available to us, diet diversity
human immune factors was responsive to the conditions and BMI scores, in this secondary data analysis.
posited in the Trivers-Willard hypothesis we would
expect them to act on the more costly immune factors,
which our results do not indicate. 4.1 | Consideration of Trivers-Willard
Although it was not possible to test the maternal buff- hypothesis in human parental investment
ering hypothesis with our data, our results, which show studies
no variation among most immune factors by maternal
condition and infant sex, suggest some indirect support The Trivers-Willard hypothesis has had a lasting impact
for the hypothesis that fluctuations in human milk on human parental investment studies (Cronk, 2007;
immune factor concentrations are shielded against nutri- Hrdy, 1987; Veller et al., 2016). There is some support for
tional and dietary fluctuations (Breakey et al., 2015; Trivers-Willard hypothesis effect in humans when look-
Fujita et al., 2019). The maternal buffering hypothesis ing at sex ratio (Cronk, 2007; Thouzeau et al., 2022).
posits that milk composition is buffered against poor However, most studies extending the Trivers-Willard
maternal condition and loss of essential nutrients due to hypothesis into investigations on post-birth investment
food uncertainty that likely occurred in the evolutionary continue to be challenged by small sample sizes
past (Fujita et al., 2019), except in extreme conditions (Cronk, 2007) and operationalization difficulties in
(Miranda et al., 1983). This buffering may occur because regard to maternal condition and parental investment
primates, including humans, synthesize milk compo- (Cronk, 2007; Veller et al., 2016). The operational vari-
nents from precursors obtained from both body stores ability in defining maternal condition is in part because
and dietary intake (Hinde & Milligan, 2011), which may “condition” is a theoretical and general term related to
explain why studies have shown some human milk com- an individual's ability to contribute to fitness (Trivers &
ponents to be resistant to short-term fluctuations in die- Willard, 1973), but in practice, condition is a multidimen-
tary intake (Fujita et al., 2019; Miranda et al., 1983; sional construct. Ideally, multiple aspects of maternal
Prentice, Prentice, et al., 1984). Therefore, the maternal condition should be simultaneously considered
buffering hypothesis may partially explain why we did (e.g., social economic status, access to resources, social
not see a wider range of relationships between maternal support, environmental and mortality risk, physical char-
condition and the immune factors in our study. acteristics [BMI, body measurements], and health condi-
Potential limitations include our inclusion criteria of tions). However, this remains challenging for many
only enrolling women who self-reported as healthy. In studies, particularly as noted above with sample size or
10 of 14 CAFFE
 ET AL.
in secondary analyses such as those presented here. Indi- infant health and milk immune composition. These find-
viduals may have high condition in some domains and ings add the investigation of human milk as a form of
low condition in others (e.g., high body mass index [BMI] biological investment and human milk composition as
and low diet diversity) and consensus on how to measure subject to parental investment tradeoffs.
a holistic ‘maternal condition’ in a complex species like
humans has yet to be and may never be obtained. Addi- AUTHOR CONTRIBUTIONS
tionally, there is no general pattern of human parental Beatrice Caffé, Courtney L. Meehan, and Aaron Black-
investment and facultative biasing of investment may well designed the study; Courtney L. Meehan, Katherine
depend on parent's local ecology (Cronk, 2007; Hrdy, Flores, Kimberly A. Lackey, Elizabeth W. Kamau-
1987; Quinn, 2013). For example, parents in one locale Mbuthia, Egidioh W. Kamundia, Debela K. Gindola,
may reduce parental investment in one domain in Samwel Mbugua, Sophie E. Moore, Gloria E. Otoo, Ros-
response to environmental or condition cues, while par- sina G. Pareja, and Lorena Ruiz collected the samples;
ents in another location may reduce investment in a dif- Juan M. Rodríguez and Lorena Ruiz designed and per-
ferent domain. Moreover, humans require multiple forms formed the laboratory analyses; Beatrice Caffé, Aaron
of parental investment to achieve reproductive success Blackwell and Courtney L. Meehan performed the data
(Quinlan et al., 2003) and receive investment from multi- analyses; Michelle K. McGuire, Courtney L. Meehan,
ple caregivers (Helfrecht et al., 2020; Meehan, 2005), fur- Mark A. McGuire, James A. Foster, Samwel Mbugua,
ther complicating the testing of Trivers-Willard Lars Bode oversaw the parent study and obtained fund-
hypothesis in human populations. While our study was ing; Beatrice Caffé, Aaron Blackwell and Courtney
able to control for some of these challenges (e.g., larger L. Meehan drafted the manuscript. All authors read, con-
sample size, cross-cultural sample), we were certainly not tributed to, and approved the final manuscript.
able to control for all (e.g., the possibility of locally rele-
vant facultative investment domains). ACKNOWLEDGMENTS
As these challenges cannot likely ever be fully or ade- This study is supported with funds from the National Sci-
quately addressed in a species as complex as humans, ence Foundation's INSPIRE Track 1 Grant: What is Nor-
perhaps, and as noted by others (Cronk, 2007; mal Milk? Sociocultural, Evolutionary, Environmental,
Quinn, 2013; Veller et al., 2016), predictions from and Microbial Aspects of Human Milk Composition
Trivers-Willard hypothesis may not be relevantly applica- (Award #1344288), National Institutes of Health NICHD
ble to post-birth human parental investment. Neverthe- R01 HD092297 and the US Department of Agriculture
less, and of concern, the Trivers-Willard hypothesis is National Institute of Food and Agriculture, Hatch project
either directly or indirectly referenced as a framework for IDA01643 and in-kind donations from Medela. We sin-
understanding variation in human milk factors, but cerely thank the Washington State University Health
rarely explicitly tested [with the exception of Quinn Equity Center for their support. Additionally, we thank
(2013) and Fujita et al. (2012)]. We argue that evidence Andrew Doel (Medical Research Council Unit, The
for the Trivers-Willard hypothesis is limited and suggest Gambia) for field supervision and logistics planning and
it does not apply as a general framework to explain varia- Alansan Sey for questionnaire administration and taking
tion in human milk immune factors or milk composition. anthropometric measurements in The Gambia; Jane Odei
We advise caution in use and skepticism of its explana- (University of Ghana) for supervising field data collection
tory power for milk factors and post-birth investment. in Ghana; Haile Belachew (Hawassa University), and Bir-
In summary, we found only limited support for the hanu Sintayehu for planning and logistics and the admin-
use of Trivers-Willard hypothesis in explaining variation istration and staff at Adare Hospital in Hawassa for
in human milk immune factors. We found only one sig- assistance with logistics in Ethiopia; Catherine
nificant relationship between maternal DDS, infant sex O. Sarange (Egerton University) for field supervision and
and concentrations of IgG; no other relationships were logistics planning and Milka W. Churuge and Minne
significant. Specifically, we found that IgG concentrations M. Gachau for recruiting, questionnaire administration,
were lower in the milk of mothers with male infants con- and taking anthropometric measurements in Kenya;
suming low diversity diets than in the milk of mothers Gisella Barbagelatta (Instituto de Investigacio
n Nutricio-
with female infants consuming low diversity diets. To our nal) for field supervision and logistics planning, Patricia
knowledge, this is the largest study to test the Trivers- Calderon (Instituto de Investigacio
n Nutricional) for
Willard hypothesis on human milk composition and to recruiting, questionnaire administration, and taking
investigate variance of IgG, IgM, and IL-10 concentra- anthropometric measurements, and Roxana Barrutia
tions within this framework. This study contributes to (Instituto de Investigacio
n Nutricional) for the manage-
our understanding of the relationship between maternal– ment and shipping of samples in Peru; Leo
nides
CAFFE
 ET AL. 11 of 14
Fern
andez, Cristina García-Carral and Irene Espinosa Brandtzaeg, P. (2003). Mucosal immunity: Integration between
(Complutense University of Madrid) for technical assis- mother and the breast-fed infant. Vaccine, 21(24), 3382–3388.
tance and expertise, and M. Ángeles Checa (Zaragoza, https://doi.org/10.1016/s0264-410x(03)00338-4
Spain), Katalina Legarra (Guernica, Spain), and Julia Breakey, A. A., Hinde, K., Valeggia, C. R., Sinofsky, A., &
Ellison, P. T. (2015). Illness in breastfeeding infants relates to
Mínguez (Huesca, Spain) for participation in the collec- concentration of lactoferrin and secretory Immunoglobulin A
tion of samples in Spain; Kirsti Kaski and Maije Sjöstrand in mother's milk. Evolution, Medicine, and Public Health, 1, 21–
(both Helsingborg Hospital) for participation in the col- 31. https://doi.org/10.1093/emph/eov002
lection of samples, questionnaire administration, and Bryan, D.-L., Hart, P. H., Forsyth, K. D., & Gibson, R. A. (2007).
anthropometric measurements in Sweden; Renee Bridge Immunomodulatory constituents of human milk change in
and Kara Sunderland (both University of California, San response to infant bronchiolitis. Pediatric Allergy and Immunology,
Diego); Janae Carrothers and Shelby Hix (Washington 18(6), 495–502. https://doi.org/10.1111/j.1399-3038.2007.00565.x
Butler, J. E., Rainard, P., Lippolis, J., Salmon, H., & Kacskovics, I.
State University) for logistics planning, recruiting, ques-
(2015). The mammary gland in mucosal and regional immu-
tionnaire administration, sample collection, and taking nity. In Mucosal Immunology (pp. 2269–2306). Elsevier. https://
anthropometric measurements in California and doi.org/10.1016/B978-0-12-415847-4.00116-6
Washington; Glenn Miller (Washington State University) Butte, N. F., & King, J. C. (2005). Energy requirements during preg-
for his expertise and critical logistic help that were nancy and lactation. Public Health Nutrition, 8(7A), 1010–1027.
needed for shipping samples and supplies worldwide. https://doi.org/10.1079/phn2005793
Castellote, C., Casillas, R., Ramírez-Santana, C., Pérez-Cano, F. J.,
CONFLICT OF INTEREST STATEMENT Castell, M., Moretones, M. G., Lo
pez-Sabater, M. C., &
Franch, A. (2011). Premature delivery influences the immuno-
The authors report no conflict of interest.
logical composition of colostrum and transitional and mature
human milk. The Journal of Nutrition, 141(6), 1181–1187.
DATA AVAILABILITY STATEMENT https://doi.org/10.3945/jn.110.133652
The data that support the findings of this study are avail- Castro, I., García-Carral, C., Furst, A., Khwajazada, S., García, J.,
able from the corresponding author upon reasonable Arroyo, R., Ruiz, L., Rodríguez, J. M., Bode, L., & Fern
andez, L.
request. (2022). Interactions between human milk oligosaccharides,
microbiota and immune factors in milk of women with and
ORCID without mastitis. Scientific Reports, 12(1), 1367. https://doi.org/
10.1038/s41598-022-05250-7
Courtney L. Meehan https://orcid.org/0000-0003-2034- CDC. (2021). Defining Adult Overweight and Obesity. Centers for
652X Disease Control and Prevention. https://www.cdc.gov/obesity/
adult/defining.html
REFERENCES Cronk, L. (2007). Boy or girl: Gender preferences from a Darwinian
Agarwal, S., Karmaus, W., Davis, S., & Gangur, V. (2011). Review: point of view. Reproductive Biomedicine Online, 15, 23–32.
Immune markers in breast milk and fetal and maternal body https://doi.org/10.1016/S1472-6483(10)60546-9
fluids: A systematic review of perinatal concentrations. Journal Czosnykowska-Łukacka, M., Lis-Kuberka, J., Kro
lak-Olejnik, B., &
of Human Lactation: Official Journal of International Lactation Orczyk-Pawiłowicz, M. (2020). Changes in human milk immu-
Consultant Association, 27(2), 171–186. https://doi.org/10.1177/ noglobulin profile during prolonged lactation. Frontiers in Pedi-
0890334410395761 atrics, 8, 428. https://doi.org/10.3389/fped.2020.00428
Amoudruz, P., Holmlund, U., Schollin, J., Sverremark- Eckart, E. K., Peck, J. D., Kharbanda, E. O., Nagel, E. M.,
Ekström, E., & Montgomery, S. M. (2009). Maternal country of Fields, D. A., & Demerath, E. W. (2021). Infant sex differences
birth and previous pregnancies are associated with breast milk in human milk intake and composition from 1- to 3-month
characteristics. Pediatric Allergy and Immunology, 20(1), 19–29. post-delivery in a healthy United States cohort. Annals of
https://doi.org/10.1111/j.1399-3038.2008.00754.x Human Biology, 48(6), 455–465. https://doi.org/10.1080/
Andreas, N. J., Kampmann, B., & Mehring Le-Doare, K. (2015). 03014460.2021.1998620
Human breast milk: A review on its composition and bioactiv- Escuder-Vieco, D., Espinosa-Martos, I., Rodríguez, J. M.,
ity. Early Human Development, 91(11), 629–635. https://doi. Fern
andez, L., & Pall
as-Alonso, C. R. (2018). Effect of HTST
org/10.1016/j.earlhumdev.2015.08.013 and holder pasteurization on the concentration of immuno-
Ballard, O., & Morrow, A. L. (2013). Human milk composition: globulins, growth factors, and hormones in donor human milk.
Nutrients and bioactive factors. Pediatric Clinics of North Amer- Frontiers in Immunology, 9, 2222. https://doi.org/10.3389/
ica, 60(1), 49–74. https://doi.org/10.1016/j.pcl.2012.10.002 fimmu.2018.02222
Böttcher, M. F., Abrahamsson, T. R., Fredriksson, M., Espinosa-Martos, I., Jiménez, E., de Andrés, J., Rodríguez-
Jakobsson, T., & Björkstén, B. (2008). Low breast milk TGF-β2 Alcal
a, L. M., Tav
arez, S., Manzano, S., Fern
andez, L.,
is induced by Lactobacillus reuteri supplementation and associ- Alonso, E., Fontecha, J., & Rodríguez, J. M. (2016). Milk and
ates with reduced risk of sensitization during infancy. Pediatric blood biomarkers associated to the clinical efficacy of a probi-
Allergy and Immunology, 19(6), 497–504. https://doi.org/10. otic for the treatment of infectious mastitis. Beneficial Microbes,
1111/j.1399-3038.2007.00687.x 7(3), 305–318. https://doi.org/10.3920/BM2015.0134
12 of 14 CAFFE
 ET AL.
Espinosa-Martos, I., Montilla, A., de Segura, A. G., Escuder, D., Kang, Y., Hurley, K. M., Ruel-Bergeron, J., Monclus, A. B.,
Bustos, G., Pall
as, C., Rodríguez, J. M., Corzo, N., & Oemcke, R., Wu, L. S. F., Mitra, M., Phuka, J., Klemm, R.,
Fern
andez, L. (2013). Bacteriological, biochemical, and immu- West, K. P., & Christian, P. (2019). Household food insecurity is
nological modifications in human colostrum after Holder pas- associated with low dietary diversity among pregnant and lac-
teurisation. Journal of Pediatric Gastroenterology and Nutrition, tating women in rural Malawi. Public Health Nutrition, 22(4),
56(5), 560–568. https://doi.org/10.1097/MPG. 697–705. https://doi.org/10.1017/S1368980018002719
0b013e31828393ed Kennedy, G., Ballard, T., & Dop, M. (2011). Guidelines for measur-
Fujita, M., Roth, E., Lo, Y.-J., Hurst, C., Vollner, J., & Kendell, A. ing household and individual dietary diversity. FAO https://
(2012). In poor families, mothers' milk is richer for daughters www.fao.org/publications/card/en/c/5aacbe39-068f-513b-b17d-
than sons: A test of Trivers-Willard hypothesis in agropastoral 1d92959654ea/
settlements in Northern Kenya. American Journal of Physical Klein, L. D., Huang, J., Quinn, E. A., Martin, M. A., Breakey, A. A.,
Anthropology, 149(1), 52–59. https://doi.org/10.1002/ajpa.22092 Gurven, M., Kaplan, H., Valeggia, C., Jasienska, G., Scelza, B.,
Fujita, M., Wander, K., Paredes Ruvalcaba, N., & Brindle, E. (2019). Lebrilla, C. B., & Hinde, K. (2018). Variation among popula-
Human milk sIgA antibody in relation to maternal nutrition tions in the immune protein composition of mother's milk
and infant vulnerability in northern Kenya. Evolution, Medi- reflects subsistence pattern. Evolution, Medicine, and Public
cine, and Public Health, 1, 201–211. https://doi.org/10.1093/ Health, 1, 230–245. https://doi.org/10.1093/emph/eoy031
emph/eoz030 Lackey, K. A., Williams, J. E., Meehan, C. L., Zachek, J. A.,
Goldman, A. S., Goldblum, R. M., & Garza, C. (1983). Immunologic Benda, E. D., Price, W. J., Foster, J. A., Sellen, D. W., Kamau-
components in human milk during the second year of lactation. Mbuthia, E. W., Kamundia, E. W., Mbugua, S., Moore, S. E.,
Acta Paediatrica Scandinavica, 72(3), 461–462. https://doi.org/ Prentice, A. M., K, D. G., Kvist, L. J., Otoo, G. E., García-
10.1111/j.1651-2227.1983.tb09748.x Carral, C., Jiménez, E., Ruiz, L., … McGuire, M. K. (2019).
Gregory, R. L., Wallace, J. P., Gfell, L. E., Marks, J., & King, B. A. What's Normal? Microbiomes in human milk and infant feces
(1997). Effect of exercise on milk immunoglobulin A. Medi- are related to each other but vary geographically: The INSPIRE
cine & Science in Sports & Exercise, 29, 1596–1601. study. Frontiers in Nutrition, 6, 45. https://doi.org/10.3389/fnut.
Groer, M., Davis, M., & Steele, K. (2004). Associations between 2019.00045
human milk SIgA and maternal immune, infectious, endocrine, Lane, A. A., McGuire, M. K., McGuire, M. A., Williams, J. E.,
and stress variables. Journal of Human Lactation, 20(2), 153– Lackey, K. A., Hagen, E. H., Kaul, A., Gindola, D.,
158. https://doi.org/10.1177/0890334404264104 Gebeyehu, D., Flores, K. E., Foster, J. A., Sellen, D. W., Kamau-
Hassiotou, F., Hepworth, A. R., Metzger, P., Tat Lai, C., Mbuthia, E. W., Kamundia, E. W., Mbugua, S., Moore, S. E.,
Trengove, N., Hartmann, P. E., & Filgueira, L. (2013). Maternal Prentice, A. M., Kvist, L. J., Otoo, G. E., … Meehan, C. L.
and infant infections stimulate a rapid leukocyte response in (2019). Household composition and the infant fecal micro-
breastmilk. Clinical & Translational Immunology, 2(4), e3. biome: The INSPIRE study. American Journal of Physical
https://doi.org/10.1038/cti.2013.1 Anthropology, 169(3), 526–539. https://doi.org/10.1002/ajpa.
Helfrecht, C., Roulette, J. W., Lane, A., Sintayehu, B., & 23843
Meehan, C. L. (2020). Life history and socioecology of infancy. Lu, M., Jiang, J., Wu, K., & Li, D. (2018). Epidermal growth factor
American Journal of Physical Anthropology, 173(4), 619–629. and transforming growth factor-α in human milk of different
https://doi.org/10.1002/ajpa.24145 lactation stages and different regions and their relationship
Hinde, K., & Milligan, L. A. (2011). Primate milk: Proximate mecha- with maternal diet. Food & Function, 9(2), 1199–1204. https://
nisms and ultimate perspectives. Evolutionary Anthropology: Issues, doi.org/10.1039/C7FO00770A
News, and Reviews, 20(1), 9–23. https://doi.org/10.1002/evan.20289 Lunn, P. G. (1994). Lactation and other metabolic loads affecting
Holmlund, U., Amoudruz, P., Johansson, M. A., Haileselassie, Y., human reproduction. Annals of the new York Academy of Sci-
Ongoiba, A., Kayentao, K., Traoré, B., Doumbo, S., Schollin, J., ences, 709, 77–85. https://doi.org/10.1111/j.1749-6632.1994.
Doumbo, O., Montgomery, S. M., & Sverremark-Ekström, E. tb30389.x
(2010). Maternal country of origin, breast milk characteristics Lynch, R., Wasielewski, H., & Cronk, L. (2018). Sexual conflict and
and potential influences on immunity in offspring. Clinical & the Trivers-Willard hypothesis: Females prefer daughters and
Experimental Immunology, 162(3), 500–509. https://doi.org/10. males prefer sons. Scientific Reports, 8(1), 15463–15413. https://
1111/j.1365-2249.2010.04275.x doi.org/10.1038/s41598-018-33650-1
Hrdy, S. (1987). Sex-biased parental investment among primates McDonald, C. M., McLean, J., Kroeun, H., Talukder, A.,
and other mammals: A critical evaluation of the Trivers- Lynd, L. D., & Green, T. J. (2015). Household food insecurity
Willard hypothesis. In R. J. Gelles & J. B. Lancaster (Eds.), and dietary diversity as correlates of maternal and child under-
Child abuse and neglect (pp. 97–147). Aldine De Gruyter. nutrition in rural Cambodia. European Journal of Clinical
Hunt, K. M., Williams, J. E., Shafii, B., Hunt, M. K., Behre, R., Nutrition, 69(2), 242–246. https://doi.org/10.1038/ejcn.2014.161
Ting, R., McGuire, M. K., & McGuire, M. A. (2013). Mastitis is McGuire, M. K., Meehan, C. L., McGuire, M. A., Williams, J. E.,
associated with increased free fatty acids, somatic cell count, Foster, J., Sellen, D. W., Kamau-Mbuthia, E. W.,
and interleukin-8 concentrations in human milk. Breastfeeding Kamundia, E. W., Mbugua, S., Moore, S. E., Prentice, A. M.,
Medicine, 8(1), 105–110. https://doi.org/10.1089/bfm.2011.0141 Kvist, L. J., Otoo, G. E., Brooker, S. L., Price, W. J., Shafii, B.,
Kalkwarf, H. J., & Specker, B. L. (2002). Bone mineral changes dur- Placek, C., Lackey, K. A., Robertson, B., … Bode, L. (2017).
ing pregnancy and lactation. Endocrine, 17(1), 49–53. https:// What's normal? Oligosaccharide concentrations and profiles in
doi.org/10.1385/ENDO:17:1:49 milk produced by healthy women vary geographically. The
CAFFE
 ET AL. 13 of 14
American Journal of Clinical Nutrition, 105(5), 1086–1100. attainment: Testing the Trivers-Willard hypothesis. Frontiers in
https://doi.org/10.3945/ajcn.116.139980 Psychology, 8, 242347. https://doi.org/10.3389/fpsyg.2017.01879
McGuire, M. K., Randall, A. Z., Seppo, A. E., Järvinen, K. M., Powe, C. E., Knott, C. D., & Conklin-Brittain, N. (2010). Infant sex
Meehan, C. L., Gindola, D., Williams, J. E., Sellen, D. W., predicts breast milk energy content. American Journal of
Kamau-Mbuthia, E. W., Kamundia, E. W., Mbugua, S., Human Biology, 22(1), 50–54. https://doi.org/10.1002/ajhb.
Moore, S. E., Prentice, A. M., Foster, J. A., Otoo, G. E., 20941
Rodríguez, J. M., Pareja, R. G., Bode, L., McGuire, M. A., & Prentice, A., Prentice, A. M., Cole, T. J., Paul, A. A., &
Campo, J. J. (2021). Multipathogen analysis of IgA and IgG Whitehead, R. G. (1984). Breast-milk antimicrobial factors of
antigen specificity for selected pathogens in milk produced by rural Gambian mothers. I. Influence of stage of lactation and
women from diverse geographical regions: The INSPIRE study. maternal plane of nutrition. Acta Paediatrica Scandinavica,
Frontiers in Immunology, 11, 614372. https://doi.org/10.3389/ 73(6), 796–802. https://doi.org/10.1111/j.1651-2227.1984.
fimmu.2020.614372 tb17778.x
Meehan, C. L. (2005). The effects of residential locality on parental Prentice, A., Prentice, A. M., & Lamb, W. H. (1985). Mastitis in
and alloparental investment among the Aka foragers of the rural Gambian mothers and the protection of the breast by milk
Central African Republic. Human Nature, 16(1), 58–80. https:// antimicrobial factors. Transactions of the Royal Society of Tropi-
doi.org/10.1007/s12110-005-1007-2 cal Medicine and Hygiene, 79(1), 90–95. https://doi.org/10.1016/
Miller, E. M. (2018). Ecological immunity of human milk: Life his- 0035-9203(85)90245-7
tory perspectives from the United States and Kenya. American Prentice, A., Watkinson, M., Prentice, A. M., Cole, T. J., &
Journal of Physical Anthropology, 167(2), 389–399. https://doi. Whitehead, R. G. (1984). Breast-milk antimicrobial factors of
org/10.1002/ajpa.23639 rural Gambian mothers. II. Influence of season and prevalence
Miranda, R., Saravia, N. G., Ackerman, R., Murphy, N., of infection. Acta Paediatrica Scandinavica, 73(6), 803–809.
Berman, S., & McMurray, D. N. (1983). Effect of maternal https://doi.org/10.1111/j.1651-2227.1984.tb17779.x
nutritional status on immunological substances in human Prentice, A., Prentice, A. M., Cole, T. J., & Whitehead, R. G. (1983).
colostrum and milk. The American Journal of Clinical Nutrition, Determinants of variations in breast milk protective factor con-
37(4), 632–640. https://doi.org/10.1093/ajcn/37.4.632 centrations of rural Gambian mothers. Archives of Disease in
Moles, L., Manzano, S., Fern
andez, L., Montilla, A., Corzo, N., Childhood, 58(7), 518–522. https://doi.org/10.1136/adc.58.7.518
Ares, S., Rodríguez, J. M., & Espinosa-Martos, I. (2015). Bacterio- Quinlan, R. J., Quinlan, M. B., & Flinn, M. V. (2003). Parental
logical, biochemical, and immunological properties of colostrum investment and age at weaning in a Caribbean village. Evolu-
and mature milk from mothers of extremely preterm infants. Jour- tion and Human Behavior, 24(1), 1–16. https://doi.org/10.1016/
nal of Pediatric Gastroenterology and Nutrition, 60(1), 120–126. S1090-5138(02)00104-6
https://doi.org/10.1097/MPG.0000000000000560 Quinn, E. A. (2013). No evidence for sex biases in milk macronutri-
Munblit, D., Treneva, M., Peroni, D. G., Colicino, S., Chow, L., ents, energy, or breastfeeding frequency in a sample of filipino
Dissanayeke, S., Abrol, P., Sheth, S., Pampura, A., Boner, A. L., mothers. American Journal of Physical Anthropology, 152, 209–
Geddes, D. T., Boyle, R. J., & Warner, J. O. (2016). Colostrum 216. https://doi.org/10.1002/ajpa.22346
and mature human milk of women from London, Moscow, and Riskin, A., Almog, M., Peri, R., Halasz, K., Srugo, I., & Kessel, A.
Verona: Determinants of immune composition. Nutrients, (2012). Changes in immunomodulatory constituents of human
8(11), 695. https://doi.org/10.3390/nu8110695 milk in response to active infection in the nursing infant. Pedi-
Na, M., Mehra, S., Christian, P., Ali, H., Shaikh, S., Shamim, A. A., atric Research, 71(2), 220–225. https://doi.org/10.1038/pr.
Labrique, A. B., Klemm, R. D., Wu, L. S., & West, K. P., Jr. 2011.34
(2016). Maternal dietary diversity decreases with household Rothman, K. J. (2008). BMI-related errors in the measurement of
food insecurity in rural Bangladesh: A longitudinal analysis. obesity. International Journal of Obesity, 32(Suppl 3), S56–S59.
The Journal of Nutrition, 146(10), 2109–2116. https://doi.org/10. https://doi.org/10.1038/ijo.2008.87
3945/jn.116.234229 Roux, M. E., McWilliams, M., Phillips-Quagliata, J. M., Weisz-
Pace, R. M., Williams, J. E., Robertson, B., Lackey, K. A., Carrington, P., & Lamm, M. E. (1977). Origin of IgA-secreting
Meehan, C. L., Price, W. J., Foster, J. A., Sellen, D. W., Kamau- plasma cells in the mammary gland. The Journal of Experimen-
Mbuthia, E. W., Kamundia, E. W., Mbugua, S., Moore, S. E., tal Medicine, 146(5), 1311–1322. https://doi.org/10.1084/jem.
Prentice, A. M., Kita, D. G., Kvist, L. J., Otoo, G. E., Ruiz, L., 146.5.1311
Rodríguez, J. M., Pareja, R. G., … McGuire, M. K. (2021). Varia- Ruiz, L., Espinosa-Martos, I., García-Carral, C., Manzano, S.,
tion in human milk composition is related to differences in McGuire, M. K., Meehan, C. L., McGuire, M. A.,
milk and infant fecal microbial communities. Microorganisms, Williams, J. E., Foster, J., Sellen, D. W., Kamau-
9(6), 1153. https://doi.org/10.3390/microorganisms9061153 Mbuthia, E. W., Kamundia, E. W., Mbugua, S., Moore, S. E.,
Peroni, D. G., Pescollderungg, L., Piacentini, G. L., Rigotti, E., Kvist, L. J., Otoo, G. E., Lackey, K. A., Flores, K.,
Maselli, M., Watschinger, K., Piazza, M., Pigozzi, R., & Pareja, R. G., … Rodríguez, J. M. (2017). What's normal?
Boner, A. L. (2010). Immune regulatory cytokines in the milk Immune profiling of human Milk from healthy women living
of lactating women from farming and urban environments. in different geographical and socioeconomic settings. Fron-
Pediatric Allergy and Immunology, 21(6), 977–982. https://doi. tiers in Immunology, 8, 696. https://doi.org/10.3389/fimmu.
org/10.1111/j.1399-3038.2010.00995.x 2017.00696
Pink, K. E., Schaman, A., & Fieder, M. (2017). Sex differences in Singh, D. R., Ghimire, S., Upadhayay, S. R., Singh, S., &
intergenerational income transmission and educational Ghimire, U. (2020). Food insecurity and dietary diversity
14 of 14 CAFFE
 ET AL.
among lactating mothers in the urban municipality in the Valeggia, C., & Ellison, P. T. (2001). Lactation, energetics, and post-
mountains of Nepal. PLoS One, 15(1), e0227873. partum fecundity. In P. T. Ellison (Ed.), Reproductive ecology
Song, S. (2018). Spending patterns of Chinese parents on children's and human evolution (pp. 85–106). Routledge.
backpacks support the Trivers-Willard hypothesis. Evolution Veller, C., Haig, D., & Nowak, M. A. (2016). The Trivers–Willard
and Human Behavior, 39(3), 336–342. https://doi.org/10.1016/j. hypothesis: Sex ratio or investment? Proceedings of the Royal
evolhumbehav.2018.02.005 Society. B, Biological Sciences, 283(1830), 20160126. https://doi.
Striker, G. A. J., Casanova, L. D., & Nagao, A. T. (2004). Influence org/10.1098/rspb.2016.0126
of type of delivery on A, G and M immunoglobulin concentra- Weaver, L. T., Arthur, H. M. L., Bunn, J. E. G., & Thomas, J. E.
tion in maternal colostrum. Jornal de Pediatria, 80(2), 123–128. (1998). Human milk IgA concentrations during the first year of
Thouzeau, V., Bollée, J., Cristia, A., & Chevallier, C. (2022). lactation. Archives of Disease in Childhood, 78(3), 235–239.
Decades of Trivers-Willard research on humans: What conclu- https://doi.org/10.1136/adc.78.3.235
sions can be drawn? BioRxiv, 2022(8), 22.504743. https://doi.
org/10.1101/2022.08.22.504743
Tomici
c, S., Johansson, G., Voor, T., Björkstén, B., Böttcher, M. F., & SUPPORTING INFORMATION
Jenmalm, M. C. (2010). Breast milk cytokine and IgA composition Additional supporting information can be found online
differ in Estonian and Swedish mothers-relationship to microbial in the Supporting Information section at the end of this
pressure and infant allergy. Pediatric Research, 68(4), 330–334. article.
https://doi.org/10.1203/PDR.0b013e3181ee049d
Trivers, R. L., & Willard, D. E. (1973). Natural selection of paren-
tal ability to vary the sex ratio offspring. Science (New York, How to cite this article: Caffé, B., Blackwell, A.,
NY), 179(4068), 90–92. https://doi.org/10.1126/science.179. Fehrenkamp, B. D., Williams, J. E., Pace, R. M.,
4068.90 Lackey, K. A., Ruiz, L., Rodríguez, J. M., McGuire,
Turin, C. G., & Ochoa, T. J. (2014). The role of maternal breast milk
M. A., Foster, J. A., Sellen, D. W., Kamau-Mbuthia,
in preventing infantile diarrhea in the developing world. Cur-
rent Tropical Medicine Reports, 1(2), 97–105. https://doi.org/10. E. W., Kamundia, E. W., Mbugua, S., Moore, S. E.,
1007/s40475-014-0015-x Prentice, A. M., Kvist, L. J., Otoo, G. E., Pareja,
Urwin, H. J., Miles, E. A., Noakes, P. S., Kremmyda, L.-S., R. G., … Meehan, C. L. (2023). Human milk
Vlachava, M., Diaper, N. D., Pérez-Cano, F. J., immune factors, maternal nutritional status, and
Godfrey, K. M., Calder, P. C., & Yaqoob, P. (2012). Salmon infant sex: The INSPIRE study. American Journal
consumption during pregnancy alters fatty acid composition of Human Biology, e23943. https://doi.org/10.1002/
and secretory IgA concentration in human breast milk. The ajhb.23943
Journal of Nutrition, 142(8), 1603–1610. https://doi.org/10.
3945/jn.112.160804