This study examined the direct genetic and indirect environmental pathways through which polygenic liability for psychopathology on internalising and externalising behaviours is transmitted from parents to offspring. Using trio genomic data, parent-reported offspring outcomes, and child-reported perspectives of parenting, we found that parental genetic predisposition for psychopathology influences adolescents’ internalising behaviours through environmental rather than direct genetic pathways—beyond the effect of shared genes.

Different to our expectations, we found no evidence that adolescents’ own genetic makeup directly influenced their internalising or externalising behaviours (i.e., child direct genetic effect). While the initial pairwise correlations showed robust associations between offspring PGS for genomic p and parent-reported behaviours of psychopathology, these associations became non-significant when parental p-PGSs were accounted for in the model.

One possibility for the absence of significant direct genetic effects is that, once environmental pathways are accounted for, genetic transmission effects may become attenuated, as they could operate through a more complex gene-environment interplay. Situational familial characteristics—such as socioeconomic status, spousal relationships, and parental social support—are candidate environmental pathways that we did not examine in this study but have been reported to moderate intergenerational transmission of psychopathology [58]. These factors may create environmental conditions that can ‘override’ the effects of a child’s alleles on their own behaviour. In other words, instead of directly observing the child’s genetic contribution to certain traits, what we see might be more strongly shaped by environmental influences. However, alternative explanations include the limited predictive power of the p-PGS, and potential suppression effects due to overlapping variance between child and parental p-PGSs.

We also observed an intriguing pattern of gene-environment interplay in the pairwise correlations, where offspring genetic predispositions were associated with both parent-reported offspring behaviour problems and child-reported parental overprotection. This may reflect evocative rGE, whereby offspring’s genetically influences behaviours elicit specific parenting responses [15, 59]. However, these associations became non-significant after controlling for parental p-PGSs, suggesting that other mechanisms may be at play. Future research using alternative genetically informed designs—such as sibling comparison or longitudinal models—will be better suited to disentangle this association.

Our findings did not support genetic nurture as a source of variance in adolescent externalising behaviours. This finding aligns with several other studies. For example, a previous study using the same cohort found no evidence of genetic nurture effects on externalising behaviours, nor environmental mediation via family dysfunction [40]. Another Dutch cohort study reported that a non-transmitted PGS for ADHD did not predict ADHD behaviours in 12-year-old offspring [23]. Similarly, a study based on the Norwegian Mother, Father, and Child Cohort Study, which examined 13 PGSs to investigate parental risk factors for conduct disorder in 14-year-old offspring, found no evidence of genetic nurture [21]. Likewise, in a study using a large UK sample, no genetic nurture effects in adolescence were found, as non-transmitted parental alleles for ADHD were not associated with ADHD liability compared to control participants [22]. To date, only one study, conducted in a high-risk sample of individuals with alcohol use disorders, has reported (a small) genetic nurture effect on externalising behaviours at age 15 [24].

A possible explanation for the absence of genetic nurture effects on externalising behaviours here is the timing of behaviour assessment [58]. Genetic nurture effects may be age-dependent, with some effects emerging more strongly earlier in development and others appearing later. Supporting this, studies have detected genetic nurture effects on externalising in both childhood and adulthood, despite their absence in adolescence. In the Norwegian cohort mentioned above, genetic nurture effects for a PGS for polygenic p—akin to the p-PGS used in our study—and for autism were associated with conduct problems and hyperactivity/inattention at age eight [30], and maternal genetic nurture effects for a neuroticism PGS were associated with ADHD traits [19]. In adulthood, genetic nurture effects have been reported for externalising traits, such as aggressive behaviour [60] and ADHD [23] in the Dutch cohort.

In contrast to externalising behaviours, we found evidence for genetic nurture for internalising behaviours for maternal—but not paternal—genetic predisposition, suggesting that these effects may be driven by mothers. This parent gender-specific effect aligns with prior research on intergenerational transmission of psychopathology, which shows maternal, rather than paternal, psychopathology is linked to adolescent internalising behaviours [61, 62], primarily through environmental mechanisms [63]. Recent studies also indicate that maternal genetic nurture can significantly influence physical and mental health [19, 64]. One plausible explanation is that mothers, often the primary caregivers, shape both the offspring’s upbringing and prenatal environment, the latter of which is critical for development and has been suggested as a critical window for genetic nurture effects. That said, offspring behaviours were reported by mothers in the absolute majority of cases (93%), which also raises the possibility of shared method variance playing a part.

Although our finding contrasts with a previous study that found no genetic nurture effects on adolescent internalising behaviours [25], this discrepancy may stem from differences in the genetic measures used. We employed a PGS for the general genetic liability to psychopathology, rather than a domain-specific PGS. In the same cohort, our analysis shows that PGS for genomic p accounts for more variance in externalising behaviour (R2 = 1.44%) than a domain-specific externalising PGS (R2 = 0.49%) used in a prior study [40], suggesting that the general genetic liability measure may serve as a more holistic PGS for studying genetic nurture effects. Supporting this, a recent study on the intergenerational transmission of childhood psychopathology [30], which used a much larger sample, constructed a similar PGS for genomic p, derived from the first principal component of PGSs for eight major psychiatric disorders. Their findings indicated that a PGS for genomic p outperformed disorder-specific PGSs in predicting children’s emotional and behavioural difficulties, and could capture maternal and paternal indirect genetic effects across both general and specific psychopathology domains, with the exception of inattention and anxiety. These findings highlight the value of taking a transdiagnostic approach and serve as a foundation for generating more specific hypotheses.

Our planned mediation analysis could not be performed due to the lack of correlation between parental p-PGSs and parenting measures, despite robust evidence for the heritability of parenting [14]. This absence of association may be due to our reliance on child-reported measures of parenting behaviour. On the one hand, this could be considered to introduce biases relating to the fact that the child’s subjective perception may be reflective of the child’s own genetic predispositions. Supporting this idea, pairwise correlations showed a significant association between offspring p-PGS and their report of parents’ behaviours. Crucially, on the other hand, this notion may be the mirror of what is seen using parent reports of parenting, that is, parent perceptions and reporting reflect their own genetics—which could also be considered “bias”. In other words, genetic nurture mediation may be more evident when parents themselves report on the environments they provide, because their own genetic makeup likely shapes both their behaviours and their perceptions of those behaviours. These possibilities underscore that, while our cross-informant design aimed to minimise shared-method bias, it does not preclude the possibility of genetically-influenced reporter perception effects. We argue that it is also possible that perception effects are not bias—rather that the nature and influence of parenting behaviours for children’s outcomes are complex, and may be partially in the genetics of the beholder. While our findings might be interpreted as indicating a minimal parental role in mediating genetic nurture, they instead highlight the complexity of genetic pathways to parenting.

The strengths of this study include the availability of genetic data capturing general genetic liability to psychopathology from both parents and their offspring, as well as phenotype data of parent-reported offspring behaviour and child-reported parenting—an approach that, to our knowledge, has not been used in previous research. However, several limitations need to be noted. First, our modest sample size limits the statistical power to detect very small genetic nurture effects. For comparison, a similar intergenerational transmission study using a trio PGS design, with a sample size ten times larger (N = 15,000), had at least 80% power to detect a small indirect genetic effect (β = 0.03) [30]. Given our smaller sample, our ability to detect such effects is likely limited, which may also affect the generalisability of our findings.

Second, our analysis was based on data from a single time point in late childhood, which may not capture effects that emerge over time or later in development. While follow-up data on a variety of topics including psychosocial adjustment, education and work, and social relationships from adulthood are available for this cohort [65], adult outcomes were beyond the scope of the present study and were not included in our analysis. Our aim was to examine proximal correlates of parenting in early adolescence, which may fade or change over time. As such, we did not attempt to model time-dependent associations. Future research could build on this work by linking the genetic nurture pathways explored in our study with later diagnostic or functional outcomes. Longitudinal designs with repeated measures are necessary to better understand how direct genetic effects and indirect genetic effects contribute to variability in psychopathology behaviours over time.

Third, while we used child-reported parenting measures to mitigate self-report bias from parents, these measures reflect the child’s subjective perception, which may not fully represent the environmental influences we intended to capture. Incorporating a range of observational methods, such as cross-rater parenting measures—including those provided by parents themselves—could provide a more nuanced understanding of gene-environment interplay. Fourth, our sample consisted of individuals of European ancestry, which limits the generalisability of our findings to more diverse populations. Lastly, expanding the range of environmental variables measured, such as additional parental factors, would provide much needed insight into the environmental mechanisms by which parental genotype shapes adolescent internalising behaviours. For instance, parental internalising behaviours could mediate this process, as developmental theories of anxiety suggest that children and adolescents may develop anxious behaviours from their parents through modelling [66], and parental criticism—particularly from mothers—has been linked to adolescent internalising behaviours [67].