Genetic and Environmental Factors in the Intergenerational Transmission of Maternal Care in Rhesus Macaques: Preliminary Findings

Abstract

Early life experiences reorganize the brain and behavior of the developing infant, often with lifelong consequences. There is perhaps no more potent developmental influence than the quality of parental care: it is an experience common to all mammals, and its effects have been observed across species. The effects of parental care can be particularly difficult to abolish, as levels of care are often perpetuated across generations. However, genetic relatedness between parents can obscure the true mechanism of transgenerational cycles of parental care, because in intact families, genes and environment are confounded. We examined the transmission of maternal care quality in biologically reared (n=21) and cross fostered (n=6) female rhesus monkeys. Interactions between female infant subjects and their mothers were observed from subjects’ birth to twelve weeks of age. Females were then observed 4-5 years later for the quality of care they displayed toward their own newborn offspring. Maternal protectiveness in the first and second generations were correlated in both biologically reared and cross-fostered females. However, other aspects of maternal care, such as aggressiveness and sensitivity, were transmitted differently depending on foster status. These data provide preliminary findings in a small sample that the intergenerational transmission of maternal care may arise from complex genetic and environmental mechanisms in rhesus monkeys.

Grapical abstract:

Introduction

Early experiences are formative. One experience of all mammalian infants share is parental care (Nowak et al., 2007; Ainsworth, 1985; Maestripieri, 2005; Pederson & Boccia, 2002; Lonstein & DeVries, 2001). Investigating the early life and demographic factors that predict parental care can help us understand, in part, familial similarity in behavioral traits.

Amongst mammals, including humans, there are consistencies in stages and general properties of care. Mothers (and fathers in biparental species) exhibit a number of species-appropriate, developmentally timed behaviors toward their infants (Ainsworth, 1969; Altman, 1980; Berman, 1990, Fairbanks, 1989; Francis et al., 1999; Hinde & Atkinson, 1970; Gubernick and Alberts, 1987; Fite and French, 2000; Wang et al., 1984; Ziegler, 2000; Ross et al., 2007; Belsky, 1979). These include tactile contact, nursing, face-to-face contact or mutual eye gaze, which give way to maintaining proximity and protection from environmental threats as the infant becomes increasingly ambulatory (Bell & Ainsworth, 1972; Ferrari et al., 2009). Most mammalian species studied to date also display natural variation in the timing and properties of parent-offspring interactions (Altman, 1980; Berman, 1980; Nowak et al., 2007; Ainsworth, 1985; Kinnally et al., 2010; McCormack et al., 2015; Caldji et al., 1998; Perkeybile and Bales, 2015). Many methods have been used to characterize the quality of parent-infant interactions, most of which include elements of parental positive and negative contact (eg, Francis et al., 1999; Berman, 1990; Weaver and de Waal, 2000; Bardi and Huffman, 2000; McKormack et al., 2015). In primate species characterized by a polygynous social system, the focus is typically on quality of maternal care, due to the fact that mothers are typically infants’ sole caregivers. Common elements of primate maternal care have included protectiveness, rejection or aggression toward infants (Fairbanks, 1989, Bardi & Huffman, 2002, Berman, 1990; Kinnally et al., 2010). Recent efforts have also been dedicated to understanding the importance of maternal sensitivity, a construct developed for humans by Mary Ainsworth (Ainsworth, 1969) and translated to rhesus monkeys (McCormack et al., 2015). This powerful construct embodies not how the mother behaves overall, but how she responds to the unique needs of the infant.

Across species, the factors that contribute to individual differences in these distinct aspects of maternal care are thought to include complex environmental and genetic processes. Evidence indicates that the quality of parental care received in childhood is highly predictive of care toward one’s own children (Belsky, 1993; Belsky et al., 2009; Spinetta & Rigler, 1972; Dodge et al., 1990; Maestripieri, 2005), although such intergenerational transmission is not inevitable (Belsky, 1980, 1984; Belsky et al., 2009). Nevertheless, the fact that the intergenerational transmission of aggression and neglect toward offspring is observed in a variety of species indicates that the phenomenon is endemic (humans: Dodge et al, 1990; macaques: Maestripieri, 2005; Berman et al., 1990; rats: Meaney, 2001; Peterson and Boccia, 2002; prairie voles: Lonstein & deVries, 2001). Elements of good quality care may also be transmitted between generations. For example, female vervets that experienced more contact with their mothers in infancy are more likely to maintain contact with their own infants (Fairbanks, 1989).

One of the challenges of this area of research is to determine whether the effects of maternal care are mediated through psychosocial mechanisms, or due to complex interactions between genetic background and behavior. Cross-fostering research with infants raised by genetically unrelated mothers demonstrates a role for environmental transmission of maternal behavior, as fostered rodents display natural variation in maternal behavior similar to the mother that raised them, whether they were foster or biological mother s(Francis et al., 1999). Seminal work in primates has shown that the experience of abusive or rejecting tendencies are often (but not always) transmitted across generations in cross-fostered infants (Maestripieri et al., 2005; Berman, 1990), but the role of gene-environment correlations in transgenerational effects of natural variation in both positive and negative aspects of maternal care is not well known.

We observed a small sample of twenty-seven female rhesus macaques as infants for the quality of care they received from their mothers. Subjects were either reared by their biological mothers (n=21) or by a genetically unrelated foster mother in a new social group (n=6). Females were then observed with their own infants 4-5 years later. If the intergenerational effects of maternal behavior are due to gene-environment correlations, we hypothesized that biologically reared females would show more similar maternal behavior to their mothers than foster infants. If the effects are environmentally mediated, both foster and biologically reared females would show similar maternal behavior to the care they received as infants.

Methods

Experimental Subjects.

Twenty-seven female infant rhesus macaques were raised with their mothers in one of seven social groups housed in half-acre outdoor enclosures at the California National Primate Research Center (CNPRC). Each field cage contained a large social group (80 – 150 members). These groups were comprised of similar social demographics, including 6-13 distinct matrilines with extended kin networks and animals of all age/sex classes. All social groups included 5-10 reproductively mature males, 25 –60 reproductively mature females, and 25-75 subadult, juvenile or infant monkeys. In first generation mothers, the number of adult female relatives in the group ranged from 5-22 members, with an average of 12.8 residing in the same enclosure. In the second generation, the number of adult female relatives ranged from 0 – 10, with an average of 3.9 female matrilineal kin in the group. Seven of our subjects resided with the mothers that raised them (6 biologically reared and 1 foster) while she was raising her own infant.

First generation mothers ranged in age from 3.5-17 years (mean = 6.5 years of age, SD = 3.57) and second generation mothers ranged from 4-6 years of age. First generation mothers were either primiparous (n = 14) or multiparous (n = 13). Second generation mothers were also either primiparous (n = 17) or multiparous (ie, third generation infants were their second born; n = 10). Six second generation females were cross-fostered and twenty one remained undisturbed with their mothers in natal groups. Pairwise relatedness among second generation females was less than 6.25%. Third generation offspring (10 females and 17 males) were reared with their second generation mothers in one of the same large half-acre outdoor enclosures, in similar conditions to second generation females when they were infants. All animal procedures were conducted in accordance with the UC Davis Institutional Animal Care and Use Committee. The research adhered to the American Society of Primatologists (ASP) Principles for the Ethical Treatment of Non Human Primates (see https://www.asp.org/society/resolutions/EthicalTreatmentOfNonHumanPrimates.cfm).

Cross-fostering.

To control for potential gene-environment correlations in the effects of maternal care, on Day 1, six mother-infant dyads were relocated indoors for standardized FOSTER procedures. Dyads were removed from outdoor enclosures on Day 1 and transported to indoor procedure rooms. Mothers were sedated (10 mg/kg ketamine) and infants were removed by trained CNPRC staff and placed on the ventrum of the new FOSTER mother who was also sedated. FOSTER mothers were always genetically unrelated females from other social groups. Dyads were monitored overnight to ensure that the FOSTER mother accepted the infant after manipulation. FOSTER mother-infant dyads were then returned to FOSTER mothers’ original social groups. Biological reared infants (BIO) remained undisturbed in their natal groups.

Mother-Infant Observations. Second and third generation infants were observed in their social groups for five minutes per observation using focal dyadic sampling (Altman, 1974), conducted one to four times weekly (range 1.25 – 4.25, average 2.27 observations per week). The number of observations per dyad did not predict any the rate at which any aspect of second-generation maternal behavior was observed (all p > .10). Observations were conducted between 0700 h and 1300h, during postnatal weeks 1-12. Observation order was rotated daily so dyads were observed at different times of day across the proscribed observation period.

Mother-infant interactions were coded using a transactional coding system, describing the overall theme of an interaction from the perspectives of the initiator and the recipient (Lyons et al., 1986; Kinnally et al., 2010, 2014). A transaction was defined as a change from one state of association that lasted three seconds or more, to a new state that was maintained for at least three seconds. Themes are defined in Table 1, and include protection, affiliation, neutral, rejection and aggression. Each dyadic interaction was characterized from the perspective of the mother and of the infant, and a theme assigned for each based on their behavior. For example, some common types of transaction types include: 1.) infant approaches mother and initiates contact, mothers initiates physical contact (which would be scored as Affiliative-Affiliative, received by the mother), 2.) Mother retrieves infant and infant grabs mother’s ventrum or back (Protective-affiliative), 3.) Infant jumped on mother, mother swats the infant to the ground (Affiliative-aggressive, received by the mother).

Table 1.

Transaction theme descriptions

Protective	Includes all behaviors intended to protect and restrict the infant’s range of movement. The defining characteristic is the action of the mother pulling the infant toward her. Protectiveness is the only transaction that infants cannot engage in.
Affiliative	Includes behaviors including prosocial physical contact, such as grooming, licking, holding, and any other non-aggressive positive touch.
Neutral	Includes non-committal behaviors. The most commons neutral transaction initiation is an approach without making physical contact. Neutral responses are those where the receiver does not react positively or negatively to an overture.
Rejecting	Includes behaviors that discourage interaction. The most common rejection theme includes walking away when other approaches. This theme can only be a response to an overture.
Aggressive	Includes physical contact aggression. Includes scratching, hitting, biting, flattening, dragging, throwing and any other physical contact that may inflict pain on the infant.

Three (one male and two female) adult raters were trained by a primatologist with experience in mother-infant interactions for at least eight two hour training sessions. Rater reliability was determined at the end of this period. by calculating whether the rater was correct in recording each aspect of the transaction (maternal theme, infant theme, and recipient) for ten sequential five-minute trials. These trials were required to include with a wide variety of transaction types or they were excluded from reliability calculations. Inter-rater reliability was 90% or better.

Rates of all transaction themes by the mother (Protective, Affiliative, Neutral, Rejecting or Aggressive) were calculated per observation period for analysis. Our maternal sensitivity measure was characterized from factor analysis of a larger dataset of 206 mother-infant dyads. Transactions in which the infant initiated social overtures toward that mother were considered in this analysis. Rates of each transaction type (continuous variable) were entered into a principal components analysis with Promax rotation. The analysis yielded five factors. We identified a factor that explained a high proportion of the total variance explained in maternal responses (17% out of 67%), and which resembled in some regards human maternal sensitivity dimensions. Higher MS scores reflect higher rates of affiliative or neutral responses to infant neutral and affiliative overtures (See Table 2 for factor loadings). This is a naturally occurring aspect of macaque maternal care that shares similarities with Ainsworth’s MS scale: high MS mothers are more accepting and symmetrical in response to infants than low MS mothers, as in humans (Ainsworth, 1969; Kinnally et al., in review) Maternal sensitivity scores were generated based on all factor loadings using the regression method.

Table 2.

Factor Analysis for Maternal Sensitivity Score

Infant (Initiator)	Mother	Factor Loading
Affiliate	Protect	−.126
Affiliate	Affiliate	.604
Affiliate	Accomodate	−.189
Affiliate	Neutral	.857
Affiliate	Resist	.075
Affiliate	Aggress	.090
Neutral	Protect	−.151
Neutral	Affiliate	−.009
Neutral	Neutral	.751
Neutral	Resist	.028
Neutral	Aggress	.075
Aggress	Aggress	.128

Maternal behavior of biological mothers was available for only 3/6 of our FOSTER subjects. We display the maternal behavior of these three FOSTER mothers and the corresponding biological mothers of our subjects, as well as group means and standard deviations, in supplementary Table 1.

Statistical Analyses

Data was examined for outliers (defined as three SD above or below the mean), and none were identified for Protectiveness, Affiliation, Neutrality or Rejection or Sensitivity. There was one outlier for aggression received and one for aggression initiated. The removal of these data points did not change the significance of the result, and so outliers were retained in the model. Because our group sizes differed, we also tested for equality of variance between groups and variance did not differ between groups for any maternal behavior measure (Levene’s test: all p > .05). To test the relationship between FOSTER maternal care and the care that females were likely to received from their biological mothers if they had not been fostered, we conducted Pearson’s correlations for each aspect of maternal care between biological and FOSTER mothers (available for only three subjects).

Backward hierarchical linear regression was conducted to determine whether individual aspects of maternal behavior received by second generation females by their own mothers predicted maternal behavior toward the third generation. Relevant demographic factors were also included in the models. Regression models included as predictors: second generation mothers’ parity, age, rank and early life foster status, third generation infant sex, all maternal behavior rates, and the foster x maternal behavior interaction term. These models find the best fit of the data, and remove predictors that do not contribute significantly to the model. We examined the potential role of social group identity, first generation mothers’ parity and age, the presence of the first generation mothers in the social group during the second generation mothers’ rearing of the third generation, and number of adult female matrilineal kin, but these factors did not predict any aspect of second generation maternal care, and were removed from the models for parsimony and to enhance power. To test the direction of foster x maternal behavior interactions, separate regression models were conducted with FOSTER and BIO groups. All models were tested using SPSS version 25. Post-hoc power achieved was calculated using GPower (Faul et al., 2007).

Results

Second generation mothers that had been BIO or FOSTER reared did not differ in any maternal care measure, either received from first generation mothers (all p > .05) nor initiated toward the third generation (p>.05). No maternal care measure was significantly correlated between our subjects’ FOSTER mothers and the biological mothers that they were cross-fostered away from (all p > .20)

Detailed regression model results are presented in Table 3. For maternal protectiveness, the overall model was significant (F (3,23) = 5.317, p = .006, adjusted R² = .332; power achieved ((1 – β error probability) = .670), where mothers who were younger, of lower rank, and who had themselves experienced higher rates of protection as infants were more protective mothers. Maternal affiliation was also predicted by our model (F (3, 26) = 8.597, p = .001, adjusted R² = .467; power achieved = .90), but only greater maternal age significantly predicted affiliation.

Table 3.

Models Predicting Second Generation Maternal Care

	B	t	p
Rate of Maternal Protectiveness
Maternal age	−.425	−2.484	.021*
Rank	.373	2.304	.031*
Maternal protectiveness received in infancy	.521	3.071	.005*
Rate of Maternal Affiliation
Maternal age	.759	4.948	.000*
Maternal parity	−.286	−1.901	.070
Maternal neutrality received in infancy	−.282	−1.888	.072
Rate of Maternal Neutrality
Maternal age	.363	1.982	.060
Maternal affiliation received in infancy	.351	1.891	.072
Maternal neutrality received in infancy	−.760	−2.570	.017*
Neutrality received x foster status	.863	3.026	.006*
Rate of Maternal Rejection
Foster status	.702	2.186	.042*
Maternal age	−.523	−3.129	.006*
Maternal protectiveness received in infancy	−.403	−2.106	.049*
Maternal neutrality received in infancy	−.463	−2.815	.011*
Maternal rejection received in infancy	1.187	2.165	.043*
Maternal aggression received in infancy	.481	2.592	.018*
Maternal rejection x foster status	−1.571	−2.558	.019*
Rate of Maternal Aggression
Foster status	−.459	−1.825	.082
Infant sex	.440	2.712	.013*
Maternal aggression received in infancy	−1.268	−2.017	.056
Maternal aggression x foster status	2.141	3.182	.004*
Maternal Sensitivity
Foster status	.526	1.897	.073
Maternal parity	−.347	−2.186	.042*
Maternal protectiveness received in infancy	−.571	−2.638	.016*
Maternal affiliation received in infancy	.470	2.669	.015*
Maternal neutrality received in infancy	−1.21	−2.476	.023*
Maternal aggression received in infancy	.615	2.920	.009*
Maternal sensitivity x foster status	1.262	2.217	.039*

Maternal neutrality in second generation mothers was also predicted significantly (F (4,22) = 3.606, p = .021, adjusted R² = .286; power achieved = .494). Older mothers and mothers that had themselves received more neutral maternal behavior were more neutral. There was an interaction between FOSTER status and neutral maternal behavior received, such that the predictive value of neutral mothering was stronger in females that had been FOSTER reared (F(1,5) = 6.863, p = .059, adjusted R² = .540), than those that had been biologically reared (F (1, 20) = .129, p = .723, adjusted R² = .00).

Maternal rejection was significantly predicted by multiple factors (F (7,19) = 5.167, p = .002, adjusted R² = .529, power achieved = .833). The effect was such that younger mothers and foster mothers were more rejecting, Additionally, females that had received less protectiveness, less neutrality, more rejection and more aggression were more rejecting. There was also a foster x rejection interaction, in which females that were FOSTER reared showed a negative relationship between rejection received and rejection initiated, but post hoc analysis showed this relationship was not significant (p > .05).

Maternal aggression was significantly predicted by two predictors (F (4, 26) = 8.547, p < .0001, adjusted R² = .537; power achieved = .939). The effect was such that male offspring in the third generation received more aggression. There was also an aggression x foster interaction such that second generation females that had been cross fostered showed a trend level relationship between aggression received and initiated ((F (1, 5) = 4.615, p = .098, adjusted R² = .42), while biologically reared females showed a significant relationship between the two factors (F (1, 20) = 7.819, p = .012, adjusted R² = .254).

Finally, the model predicting maternal sensitivity was also significant (F (7,26) = 3.691, p = .011, adjusted R² = .42; power achieved = .622). Maternal sensitivity was higher in mothers that were younger, that themselves has received less protectiveness, less neutrality, more aggression and more affiliation. There was also an interaction between maternal sensitivity received and foster status, such that females that had been fostered as infants showed a significant positive relationship between maternal sensitivity received in infancy and sensitivity directed toward third generation offspring (F (1,5) = 12.097, p = .025, adjusted R² = .689) while biologically reared females showed no relationship ((F (1,20) = .031, p = .862, adjusted R² = .00).

Discussion

Early environmental influences like maternal care are potent predictors of infant neurobehavioral development (Bowlby, 1951). We found that natural variation in the quality of care that a female monkey received from her mother early in life was somewhat associated with the quality of care she displayed toward her own offspring in adulthood, consistent with data in macaques and other species (Fairbanks 1989; Curley et al., 2008; Maestripieri et al., 2005; Francis et al., 1989; Berman, 1990). While maternal protectiveness was transmitted similarly to the second generation of both biologically reared and cross-fostered females, the intergenerational effects of other aspects of early maternal care, like aggressiveness, neutrality, and sensitivity, depended on FOSTER status. These preliminary data from a small longitudinal sample of macaques suggest that that intergenerational transmission of maternal care depends on complex environmental and genetic factors, but more investigation is clearly warranted.

First, some key demographic features predicted aspects or mothering in the second generation. We found that younger mothers were more protective, less affiliative, and more rejecting of infants. The age range was quite narrow, spanning age 4-6, and so the effect may not persist with a wider age range, but this effect likely reflects the lack of experience younger mothers had with infants. Lower ranked mothers were also more protective, likely reflecting a desire to protect infants from higher ranked individuals who could harm the infant. Maternal parity influenced only one aspect of maternal behavior, maternal sensitivity. Mothers who had previously raised an infant were more sensitive, again, likely due to their experience. Unexpectedly, we observed that our subjects were more likely to be aggressive toward male infants. This is inconsistent with our previous findings in a larger sample from this population, however, and may be an artifact (Kinnally et al., 2010).

Independent of these demographic factors, we observed that maternal protectiveness received in infancy significantly predicted maternal protectiveness directed toward our subjects’ own offspring. This is consistent with prior data in non-human primates: the amount of time spent in ventro-ventral contact with infants was conserved across generations in female vervets, for example (Fairbanks, 1989). Our model shows that key demographic factors like age and social rank predicted protectiveness as well, but these factors did not explain the transgenerational effects of early protectiveness. These variables each accounted for separate variance in explaining maternal protectiveness. The transgenerational similarity in protectiveness was observerd in both FOSTER and biologically reared infants, suggesting that maternal protectiveness may be acquired through experience, rather than an inherited genetic background for protectiveness. Maternal rejection was also predicted by the experience of rejection in childhood, consistent with a large and seminal study conducted in macaques (Berman, 1990), but only when all other maternal behaviors were included in the model. This may suggest that rejection is one part of a larger number of experiences that are linked with later rejecting behavior toward offspring. We also show that maternal aggressiveness in infancy predicted higher rates of aggression toward third generation infants. Unlike protectiveness, however, these effects were stronger in biologically reared females, although there was a trend level positive association in FOSTER infants. These data preliminarily suggest that gene-environment correlations may enhance the intergenerational transmission of maternal aggression. These data may at first appear to be inconsistent with seminal work on macaque maternal abusiveness that shows that extreme maternal aggression depends on exposure, rather than genetic relatedness (Maestripieri, 2005). However, our data suggests that biologically reared infants are more likely to mimic the degree of maternal aggression received (high or low) when there may be a genetic predisposition to one or the other.

In contrast, unexpectedly, two aspects of maternal care were transmitted with more fidelity in FOSTER reared females than biologically reared females. FOSTER mothers that received more neutral transactions from their own mothers, and experienced more maternal sensitivity, displayed similar profiles toward their own infants. In contrast, there was little relationship on these axes in biologically reared females. Since the FOSTER sample is small, it is possible that this finding is an artifact. However, these data are somewhat consistent with a developmental hypothesis has been advanced in the last decade. It has been suggested that stress early in development may sensitize infants to their environments (Pluess and Belsky, 2011). A recent experimental study confirmed that prenatal stress exaggerated the developmental consequences of early parental care (Hartman et al., 2018). It is possible that the process of cross-fostering sensitized our infants to their environments, either due to the stress of one-time fostering procedures or longer-term awareness that they are not related to their FOSTER mother. As a result, FOSTER infants may have been more vigilant in the early postnatal period and more likely to adopt their FOSTER mothers’ behavior. Our data somewhat supports this theory, but the negative relationship we observed between rejection received and displayed in FOSTER females is inconsistent with this theory. Also inconsistent is our finding that the transgenerational effects of early aggressiveness, protectiveness or affiliation were not amplified by cross-fostering. Our post-hoc power analyses suggest that we may not have been adequately powered to test whether protectiveness was amplified in FOSTERs, but the other analyses in question achieved power greater than .90, enough to detect the interaction if present. Future studies will seek to replicate these findings in a larger sample to determine whether FOSTER infants are selectively sensitized to aspects of their early environment.

A general theory for the mechanisms of transgenerational effects of maternal care is that infants are modeling the behavior they experienced or observed between their mothers and siblings (Berman et al., 1990; Fairbanks, 1989). Alternatively, one might argue that the quality of early experiences shape the psychological makeup of the individual, making certain types of maternal behavior more or less likely (Troisi and D’Amato, 1994; McCormack et al., 2009; Maestripieri et al., 2006; Maestripieri et al., 2007; Sanchez et al., 2010). For maternal rejection and maternal sensitivity, we observed that multiple axes of maternal behavior received predicted maternal behavior in adulthood. For example, rejecting behavior in adulthood was predicted by the experience of lower protectiveness, lower neutrality and higher aggression from their own mothers. Maternal sensitivity was predicted by less protectiveness and neutrality and more affiliation and aggression experienced in infancy. These additive contributions of different aspects of early maternal care toward later maternal care suggest that females did not strictly model each type of maternal behavior they experienced. This is consistent with data in rodents that shows that maternal care programs psychological traits like anxiety early in life, which may then predict maternal behavior in adulthood (Francis et al., 1999).

However, a limitation of this study cannot speak to the lifespan psychological or biological mechanisms of effect in our dataset. We know that this cycle may be mediated in part by early maternal programming of anxiety related traits, as we and others have shown (Kinnally, 2010; Caldji et al., 1998; McCormack et al., 2006; Howell et al., 2017). Early biological effects of stress, like epigenomic plasticity, may also play a role as has been demonstrated in rats (Weaver et al., 2004). In macaques, we have previously observed that higher protectiveness and lower aggressiveness predicts higher DNA methylation in an important stress-regulation gene, the serotonin transporter (Kinnally, 2014). Future studies will examine whether early epigenetic development plays a role in the transgenerational effects of maternal care in macaques.

Another important limitation of this study was the small sample size. Only twenty-seven females were observed longitudinally, in infancy and again as adults. Though small, this longitudinal dataset is valuable because female macaques typically become reproductively mature at four years of age, meaning that repeating this study would take five years or more. Another limitation is that we have not eliminated the possibility that gene-environment correlations explain transgenerational similarity in maternal behaviors that are similarly transmitted in FOSTER and biological pairs. This is not a significant problem for interpreting the transmission of aggressiveness (where we already conclude a gene-environment correlation plays a role) or neutrality and sensitivity (where we do not see any intergenerational similarity in biological pairs, when gene-environment correlations are certainly present). But for maternal protectiveness and rejection, because we did not observe the maternal behavior of 3/6 of the biological mothers of our FOSTER subjects, it is possible that by chance, biological mothers would have shown identical behavior to the FOSTER mother. In this case, our FOSTER subjects would have coincidentally experienced a genetic background linked with the maternal behavior they received from the FOSTER mother. It seems unlikely, because in the 3/6 biological mothers of FOSTER infants we have such data for (See Supplementary Table 1), there were no statistically significant correlations between FOSTER and biological mothers’ care. Closer investigation shows that while one of the three biological mothers shows similar (but not identical) patterns of behavior, in the other two pairs, average maternal behavior scores differ as much as a standard deviation between biological and FOSTER mothers of our subjects. Thus, we do not believe transgenerational similarity in protectiveness or rejection relies on gene-environment correlations. Nevertheless, our results from a smaller sample must be replicated and expanded before firm conclusions can be drawn.

In conclusion, we present preliminary evidence in a small sample that some, but not all, aspects of maternal care is transmitted to females early in development, but through complex mechanisms that are unique for each behavior. Some, like aggression, may be more potent when there is a genetic predisposition of aggressive (or non-aggressive) mothering, while some, like protectiveness, appear to be contingent on experience. Intriguingly, other aspects of maternal behavior may be transmitted with more fidelity following the mild stress of cross-fostering to and rearing by a genetically unrelated mother. Future studies will examine the mechanisms of gene-environment correlations and environmental sensitization on neurobehavioral and health consequences for the next generation.