Advanced paternal age is associated with reduced reproductive success and offspring racing performance in Australian Thoroughbred racehorses

Abstract

Reproductive ageing influences fertility and offspring fitness across mammals. In humans and model organisms, paternal age has been associated with changes in sperm quality and genomic integrity, including oxidative stress, DNA damage, and epigenetic alterations; however, these mechanisms have not been directly examined in large equine populations. Thoroughbred racehorses provide a system to investigate paternal age effects at scale using detailed breeding records. In this retrospective observational study, with data from 35 retired stallions and 33,546 progeny, we assessed associations between sire age at conception, reproductive outcomes and progeny racing performance using generalized linear mixed-effects models. Progeny conceived when stallions were ≤ 8 years were more likely to race (odds ratio [OR] = 2.21, 95% CI 1.93–2.52), place in a stakes race (OR = 3.71, 95% CI 2.74–5.04), and win a stakes race (OR = 3.58, 95% CI 2.43–5.29) than those conceived at ≥ 19 years. These findings demonstrate population-level associations between paternal age and reproductive outcomes.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-59342-9.

Introduction

An increase in the average age of first-time parents in recent decades has prompted investigation into the effects of parental age on embryo development and offspring health in humans and mouse models^1–3. Advancing paternal age has been associated with an elevated risk of numerous offspring pathologies, including cancer ^4–6, cardiovascular disease⁶, and neurodevelopmental or neurodegenerative disorders⁷. These associations are thought to arise from multiple age-related processes in the male germline, including the accumulation of replication errors during repeated divisions of spermatogonial stem cells, clonal expansion of mutant lineages^8,9, and post-meiotic mechanisms including oxidative damage to sperm DNA^10,11 and epigenetic effects^12–14.

Thoroughbred horses often have long reproductive careers after competition. Unlike other domestic species, where males typically cease breeding at much younger ages (e.g., up to 5 years of age in sheep^15,16, and 8 years of age in cattle¹⁵), many stallions remain active breeders well into their late teens or early twenties¹⁷—the equivalent to 50 years or older in humans^18,19. Despite this extended breeding lifespan, relatively little research has examined the effects of stallion ageing on germ cell quality or subsequent offspring health and performance. Evidence from humans and other mammalian species indicates that advancing paternal age is associated with increased oxidative stress and DNA damage in spermatozoa^1,7,11,20, although comparable data in the equine species remains limited.

The influence of paternal age extends beyond offspring health to physical performance, a factor of critical importance to the performance-driven Thoroughbred industry. A further complexity in assessing the impact of paternal age arises from the time lag inherent to Thoroughbred breeding. The commercial value of a breeding stallion depends on how well his progeny perform on the racetrack. Because it takes about 3 years from conception through to a racetrack debut, and then additional time to see how multiple cohorts of a stallion’s progeny perform, it typically takes around 5 years from his first breeding season before breeders can confidently judge whether he is likely to be a highly valuable sire. This delay increases the duration over which age-related changes in the male germline may occur. Evidence from other species suggests that such changes can include increased oxidative stress, DNA damage, and epigenetic alterations in sperm, although their extent and consequences in stallions remain poorly understood. Importantly, the consequences of such damage may not become evident until long after conception, and in some cases not until after a stallion’s breeding career has ended. As a result, retrospective analyses of progeny performance across a stallion population provides the only practical approach to evaluating the effects of paternal age on progeny outcomes. To date, no such analysis has been conducted on the Australian Thoroughbred population.

While previous studies in Thoroughbred horses have largely examined the heritability and genetic determinants of performance traits^21–24 and a small number have considered parental age effects, the impact of paternal age on broader measures of progeny racing success remains underexplored. The extensive performance records available, coupled with detailed pedigree documentation and a high incidence of late-life breeding in this industry, provide a unique opportunity to investigate these effects at scale. In this system, racing performance can be used as a measurable proxy for economically and functionally relevant offspring outcomes in a long-lived mammalian population. By analysing a defined subset of stallions and their progeny, this study aims to determine whether advancing paternal age influences reproductive and performance outcomes.

Results

Inclusion of progeny year of birth as a random effect did not materially alter the magnitude or direction of the estimated sire age effects across any performance outcome. Sire age remained a significant predictor of progeny performance, indicating that observed associations were not attributable to temporal variation among foal birth cohorts.

Seasonal fertility

Seasonal fertility declined with increasing sire age (Fig. 1). Estimated marginal means differed significantly among age groups (P < 0.001): stallions ≤ 8 years had the highest fertility (76.5%; 95% CI: 74.4–78.5), similar to the 9–13 year group (75.4%; 95% CI: 73.4–77.3), while fertility was lower in stallions aged 14–18 years (72.4%; 95% CI: 70.3–74.4) and lowest in those ≥ 19 years (66.8%; 95% CI: 63.7–69.6). Sire age significantly affected seasonal fertility (likelihood ratio test [LRT] χ²(3) = 31.922, P = 5.435 × 10^-07). Pairwise comparisons indicated higher odds of producing a live foal for stallions ≤ 8 years compared with those ≥ 19 years (OR = 1.62, 95% CI: 1.40–1.87), while the comparison between ≤ 8 and 9–13 years was not significant. Descriptive statistics and full pairwise odds ratios are provided in Supplementary Tables S4–S5.

Fig. 1.

The effect of increasing paternal age at conception on reproductive outcomes. (a) Seasonal fertility as a factor of stallion age, and (b) the percentage of foals who commenced a racing career (‘runners’) as a factor of stallion age. Red triangles show estimated marginal means; blue error bars are 95% confidence intervals and black dots are raw data.

Race runners

Only a proportion of foals ultimately become racehorses, as factors such as injury, physical ability, behaviour, or management decisions prevent some from starting in a race^25,26. Racing records were used to model the likelihood that individual progeny would successfully commence a racing career (e.g., compete in at least one race). The proportion of progeny that started in at least one race declined with increasing sire age (Fig. 1). Estimated marginal means differed significantly among sire age categories (P < 0.0001), with stallions ≤ 8 years producing the highest proportion of runners (81.2%; 95% CI: 79.1–83.1) and stallions ≥ 19 years the lowest (66.7%; 95% CI: 62.8–69.4). Intermediate values were observed in the 9–13 year (76.9%; 95% CI: 74.7–79.0) and 14–18 year (72.0%; 95% CI: 69.4–74.5) groups. Sire age significantly affected the likelihood of producing a race runner ((LRT) χ²(3) = 47.174, P ≤ 3.191 × 10⁻¹⁰). Pairwise comparisons showed higher odds of producing a runner for stallions ≤ 8 years compared with those ≥ 19 years (OR = 2.21, 95% CI: 1.93–2.52). Across all pairwise comparisons, younger sires consistently exhibited higher odds than older sires. Descriptive statistics and full pairwise odds ratios are provided in Supplementary Tables S4–S5.

Race winners

The proportion of progeny that won at least one race declined with increasing sire age (Fig. 2). Estimated marginal means differed significantly among sire age categories (P < 0.001), with stallions ≤ 8 years producing the highest proportion of winners (71.2%; 95% CI: 68.5–73.8) and stallions ≥ 19 years the lowest (57.2%; 95% CI: 53.1–61.1). Intermediate values were observed in the 9–13 year (66.9%; 95% CI: 64.2–69.6) and 14–18 year (62.7%; 95% CI: 59.3–65.9) groups. Sire age significantly affected the likelihood of producing a race winner (LRT χ²(3) = 27.67, P ≤ 4.26 × 10⁻⁶), with higher odds for stallions ≤ 8 years compared with those ≥ 19 years (OR = 1.85, 95% CI: 1.57–2.19). Across all pairwise comparisons, younger sires consistently exhibited higher odds than older sires. Descriptive statistics and full pairwise odds ratios are provided in Supplementary Tables S4–S5.

Stakes performers

The proportion of progeny achieving a stakes performance (e.g., came 1st, 2nd or 3rd in a stakes race) declined with increasing sire age (Fig. 2). Estimated marginal means differed significantly among sire age categories (P < 0.0001), with stallions ≤ 8 years producing the highest proportion of stakes performers (8.1%; 95% CI: 6.4–10.2) and stallions ≥ 19 years the lowest (2.3%; 95% CI: 1.6–3.3). Intermediate values were observed in the 9–13 year (6.2%; 95% CI: 4.9–7.8) and 14–18 year (4.1%; 95% CI: 3.2–5.2) groups. Sire age significantly affected the likelihood of producing a stakes performer ((LRT) χ2(3) = 57.731, P =  < 1.794 × 10^–12), with higher odds for stallions ≤ 8 years compared with those ≥ 19 years (OR = 3.71, 95% CI 2.74–5.04). Across all pairwise comparisons, younger sires consistently exhibited higher odds than older sires. Descriptive statistics and full pairwise odds ratios are provided in Supplementary Tables S4–S5.

Fig. 2.

The effect of increasing paternal age at conception on progeny racing performance. (a) The proportion of race runners sired by each stallion age group that became race winners, (b) the proportion that achieved a placing in a stakes race (1st–3rd), (c) the proportion that won a stakes race, and (d) the proportion that won a group race. Red triangles indicate estimated marginal means; blue error bars show 95% confidence intervals and black dots represent raw data.

Stakes winners

The proportion of progeny achieving a stakes win declined with increasing sire age (Fig. 2). Estimated marginal means differed significantly among sire age categories (P < 0.01), with stallions ≤ 8 years producing the highest proportion of stakes winners (4.3%; 95% CI: 3.5–5.5) and stallions ≥ 19 years the lowest (1.3%; 95% CI: 0.8–1.9). Intermediate values were observed in the 9–13 year (3.2%; 95% CI: 2.5–4.1) and 14–18 year (2.2%; 95% CI: 1.7–2.9) groups. Sire age significantly affected the likelihood of producing a stakes winner ((LRT) χ2(3) = 55.281, P =  < 5.98 × 10^–12), with higher odds for stallions ≤ 8 years compared with those ≥ 19 years (OR = 3.58, 95% CI 2.43–5.29). Across all pairwise comparisons, younger sires consistently exhibited higher odds than older sires. Descriptive statistics and full pairwise odds ratios are provided in Supplementary Tables S4–S5.

Group winners

The proportion of progeny winning at Group-level races decreased with advancing sire age (Fig. 2). Estimated marginal means differed significantly among sire age categories (P < 0.05), with stallions ≤ 8 years producing the highest proportion of group winners (2.0%; 95% CI: 1.5–2.7) and stallions ≥ 19 years the lowest (0.5%; 95% CI: 0.2–0.9). Intermediate values were observed in the 9–13 year (1.4%; 95% CI: 1.0–1.9) and 14–18 year (1.1%; 95% CI: 0.8–1.5) groups. Sire age significantly affected the likelihood of producing a group winner ((LRT) χ2(3) = 37.572, P = 3.482 × 10^–8), with higher odds for stallions ≤ 8 years compared with those ≥ 19 years (OR = 4.40, 95% CI 2.38–8.13). Across all pairwise comparisons, younger sires consistently exhibited higher odds than older sires. Descriptive statistics and full pairwise odds ratios are provided in Supplementary Tables S4–S5.

Maternal age

Maternal age increased modestly with advancing sire age, with estimated marginal means ranging from 9.3 years (95% CI: 9.1–9.5) for stallions ≤ 8 years to 10.1 years (95% CI: 9.9–10.3) for stallions ≥ 19 years. Median mare age remained 9 years across all sire age categories, and ages were highly variable within groups (range 2–26 years), indicating that the small shift in mean age is unlikely to meaningfully contribute to the observed declines in fertility or progeny performance. The number of mares bred per sire age category and mare age summary statistics are provided in Supplementary Table S3.

Continuous sire-age analyses

To complement the primary analyses based on sire-age categories, sire age was also modelled as a continuous variable. Continuous sire-age analyses yielded similar trends to those observed in the categorical analyses, with progressive declines in seasonal fertility and progeny performance outcomes across the sire-age range (Supplementary Figure S6).

Discussion

This retrospective observational study identified population-level associations between advancing paternal age and both reproductive outcomes and progeny racing performance in Thoroughbred horses. Increasing paternal age was associated with decreases in both median and mean values for all variables examined, indicating a consistent age-related decline in sire reproductive efficiency and progeny performance. This pattern was evident across measures of seasonal fertility, runners-to-foals ratios, stakes performers ratios, and the production of stakes- and group-winners, suggesting that advancing sire age is associated with reduced likelihood of producing a live foal and of producing offspring that race, win, and achieve elite-level success. However, it should be noted that there will always be a cohort of stallions for whom this pattern does not apply. These findings indicate a population-level association between paternal age and reproductive and performance outcomes in the Australian Thoroughbred population.

Seasonal fertility showed an age-associated decline, with stallions aged ≥ 19 years exhibiting significantly lower fertility than all younger age groups (Supplementary Table S4). While fertility rates were not significantly different between stallions aged ≤ 8 and 9–13 years, a significant reduction was observed from 14 years onwards. These findings align with previous reports of declining reproductive performance in ageing stallions due to testicular degeneration^17,27. Furthermore, sperm quality in horses follows a bell-shaped curve with age. Sperm quality peaks around 12 years of age but declines after 13 years^17,28,29, with older stallions producing less semen and lower-quality sperm^29,30, supporting the statistically significant decline in fertility seen in these stallions beyond 13 years of age. During the present study, odds ratio analyses supported these previous findings, as younger sires consistently had greater odds of producing a live foal than older sires, particularly when contrasted against stallions aged ≥ 19 years.

Beyond fertility, advancing sire age was associated with substantial differences in progeny racing performance. The probability that progeny would ever commence a racing career declined steadily but significantly with increasing paternal age, even after adjusting the model to account for unequal numbers of foals produced per sire (Supplementary Table S4). This decline was accompanied by increasing variability among older stallions, suggesting greater inter-stallion differences in ageing trajectories. While on average younger sires were more likely to produce progeny that would make it to the racetrack (Supplementary Table S4), the magnitude of decline varied between individuals, indicating that sire-specific factors may influence the rate at which reproductive and performance outcomes change with age.

Paternal ageing affected not only the likelihood that progeny would commence a racing career, but also their success on the racetrack, with the proportion of progeny to win at least one race, achieve a stakes placing, or win a stakes or group race all decreasing progressively with increasing sire age at conception (Supplementary Table S4). Odds ratio analyses consistently showed that, for every comparison, the younger of the two sire age groups had higher odds of producing winning, stakes-performing, or stakes- or group-winning progeny (Supplementary Table S5). Notably, stallions aged ≥ 19 years showed the lowest odds across all performance outcomes, while stallions aged ≤ 8 years consistently exhibited the highest odds of producing elite progeny.

The decline in mean and median percentages highlights a progressive reduction in reproductive success and progeny performance that could be associated with advancing paternal age. Stallion identity was included as a random effect in generalized linear mixed-effects models to account for repeated measures and inherent differences among sires, recognising that individuals vary in baseline fertility and the rate of age-related decline. Differences in susceptibility to age-related decline may reflect variation in genetic background, management practices, cumulative breeding load, or environmental stressors, and warrant further investigation. Consequently, direct comparisons between stallions of the same chronological age may be misleading; biologically meaningful comparisons are best made within stallions across their own breeding lifespan. Despite this individual variation, the large population scale of the study reveals a clear, consistent pattern of age-associated decline. Similarly, the inclusion of progeny year of birth is particularly important given potential changes in training practices, race quality, and industry structure over time. Although cohort-level variation in performance was observed (Supplementary Table S8), sire age was not strongly confounded by progeny birth year due to overlapping sire age distributions across foal cohorts within the study population. However, as this study is observational and restricted to a defined cohort of stallions, residual confounding cannot be fully excluded.

Maternal factors represent an important source of variation in this study. A small increase in mean mare age was observed across sire age categories (9.3–10.1 years), suggesting that changes in mare age structure are unlikely to account for the observed declines (Supplementary Table S3). Although mares inseminated at 2 years of age were included to reflect commercial breeding practice, they may contribute to residual heterogeneity. This subgroup could not be independently evaluated due to incomplete linkage between studbook progeny identifiers and individual performance records. In addition, mare “quality” variables (e.g. genetic merit, performance history, and reproductive success) were not consistently available within this historical dataset and could not be incorporated into the models. These limitations highlight the absence of detailed mare-level covariates as a potential source of unexplained variation in the observed outcomes. The mixed-model framework allowed partial control of unmeasured environmental and management variation—including diet, housing, seasonal weather, exercise, and veterinary treatment—that are rarely available in large-scale breeding datasets.

Although 35 stallions were included per age category (28 in the ≥ 19-year group), individual sires exert substantial population-level influence due to the large number of mares bred annually. Unlike mares, which are limited to producing a single foal per year, stallions may breed well over one hundred mares per season, amplifying the downstream consequences of age-related declines in fertility and progeny performance. The reduced representation of stallions in the ≥ 19-year category reflects industry attrition with advancing age, whereby only the most commercially successful sires typically remain active breeders. This introduces a positive selection bias within the oldest cohort, such that these stallions represent a genetically and commercially elite subset. The persistence of performance decline despite this bias suggests that the observed associations are unlikely to be explained solely by diminished genetic merit and are consistent with age-related biological changes.

As this is a retrospective observational study, the present findings can only be interpreted as associations, and any mechanistic explanations remain speculative and derived from previous literature. Age-related declines in progeny performance observed in this study are consistent with patterns reported in other species, where age-related changes in the male germline have been proposed to contribute to similar outcomes. In horses, previous studies have reported an age-associated decline in semen quality and fertility, including a reduction in sperm motility, normal morphology, and semen volume. These findings are broadly consistent with the seasonal fertility decline observed here. However, fewer studies have examined downstream offspring performance in equine populations, limiting direct comparison. As causal mechanisms cannot be directly inferred, several age-related processes may plausibly contribute to the observed patterns. One class of mechanisms operates prior to meiosis, during the lifelong mitotic divisions of spermatogonial stem cells. In humans, paternal age is associated with increasing numbers of de novo mutations transmitted to offspring³¹, consistent with the cumulative replication errors that can arise as germline stem cells divide throughout life. Recent large-scale sequencing studies further indicate that mutation patterns in the male germline can reflect clonal expansion and positive selection of spermatogonial lineages, resulting in age-dependent shifts in the genetic composition of sperm populations^8,9.

A second class of mechanisms may operate after meiosis, affecting the integrity of mature spermatozoa. Ageing has been associated with increased susceptibility of sperm DNA to oxidative damage and altered epigenetic regulation, processes that can compromise fertilisation success and early embryonic development^4,11–13. Stallion spermatozoa are considered particularly vulnerable to oxidative stress due to their high polyunsaturated fatty-acid content and limited antioxidant defences³², and increased DNA damage has been reported in ageing stallions^27–30. Although the present study does not examine molecular mechanisms directly, the observed associations between advancing sire age and declines in fertility and progeny performance are compatible with several biological hypotheses. These include both pre-meiotic germline mutation processes and post-meiotic damage to sperm DNA or epigenetic regulation. However, these mechanisms remain speculative in the absence of direct molecular data.

Our results demonstrate population-level associations between paternal age and multiple reproductive and performance traits. Although individual stallions differed in the rate and magnitude of decline, the population-level reductions in both mean and median performance metrics indicate a systematic population-level pattern rather than isolated reproductive failure. These findings extend on previous reports of age-related declines in stallion fertility and semen quality by demonstrating corresponding associations with downstream progeny performance. In other species, similar patterns have been proposed to involve cumulative oxidative damage, mutation accumulation, and epigenetic changes within the male germline, although these mechanisms were not assessed in the present study.

These results are relevant to broader questions in evolutionary and population biology concerning how paternal age shapes variation in progeny performance. In species where males reproduce across extended portions of the lifespan, individuals can contribute offspring produced at markedly different ages, generating intra-male variation in reproductive outcomes. Such patterns may influence the distribution of reproductive success within populations and affect how reproductive longevity interacts with offspring quality. Empirical evidence for these dynamics in large mammals remains limited because long-term reproductive and performance records are rarely available at sufficient scale. The extensive pedigree and racing records available for Thoroughbred horses provide a rare opportunity to examine these patterns across large populations. The consistent age-associated decline in fertility and progeny performance observed in this study demonstrate population-level associations between paternal age and offspring outcomes, highlighting the value of the Thoroughbred horse as a model for investigating male reproductive ageing in long-lived mammals.

Conclusion

Advancing paternal age was associated with consistent decline in both reproductive success and progeny racing performance across the Australian Thoroughbred population assessed in this study. Stallions produced fewer live foals as they aged, and offspring conceived later in a sire’s breeding career were less likely to race, win races, or achieve elite performance outcomes. These findings demonstrate a population-level association between paternal age and reproductive outcomes in a long-lived mammalian species. The observed patterns are consistent with, but do not demonstrate, age-related changes in sperm quality, DNA integrity, epigenetic regulation, or other germline processes. Molecular mechanisms were not directly assessed in this study and therefore remain speculative. The scale and longitudinal depth of this cohort position the Thoroughbred as a useful cross-species model for investigating male germline ageing, although the underlying biological mechanisms remain to be determined and should be the focus of further studies.

Methods

This study was a retrospective observational analysis of studbook breeding and performance records for Australian Thoroughbred stallions.

Stallion selection and data collection

A comprehensive list comprising 557 stallions who were actively breeding in Australia from 1990 to 2001 was compiled from the annual Stallions publication (https://www.stallions.com.au/). Inclusion criteria were applied to ensure that the study focused on commercially significant sires. These criteria were: commenced breeding in 1990 or later; bred for a minimum of 10 Australian breeding seasons (from 1st September to 31st May inclusive); bred an average of 50 mares each year; bred until at least 15 years of age; produced at least 650 foals conceived in Australia; were retired from stud, or deceased at the time of data collection; and their youngest progeny ≥ 5 years of age to allow sufficient time for evaluation of racing outcomes. After applying these criteria, the dataset was reduced to 35 stallions.

Breeding data for each stallion were obtained manually from the Australian Studbook online database (https://www.studbook.org.au/). For each year at stud in Australia, the following additional data were collected: total number of breeding events (n = 48,313), total number of unique mares bred (n = 24,180), age of each mare at the time of breeding, and the number of live foals produced (n = 33,546). Using these variables, annual seasonal fertility was calculated for each stallion as the number of live foals produced divided by the number of unique mares recorded in the Australian Stud Book as having been bred to that stallion during the corresponding breeding season (live foals per mare bred). The Australian breeding season is defined as 1 September to 31 May, with matings recorded outside this period assigned to the corresponding breeding year according to Stud Book rules. The number of mares bred varied across stallion age groups, reflecting differences in breeding activity and cohort attrition with advancing age.

Stallion age was categorised into four groups (≤ 8, 9–13, 14–18, and ≥ 19 years) representing young, mature, aged, and geriatric stages of the breeding career, respectively. These cut-offs were selected to reflect biologically meaningful stages of reproductive ageing in stallions and to facilitate interpretation of effects across distinct breeding-life phases. This approach is consistent with established evidence of progressive declines in stallion reproductive function, including reductions in semen quality and fertility from approximately 13–15 years of age onwards, as described in equine reproductive literature^17,27–29. Categorisation of 5-year intervals also ensured sufficient sample sizes within each group for robust statistical comparison of outcomes.

The study design involved repeated measurements of the same stallions across multiple sire age categories throughout their breeding careers. Accordingly, stallions contributed data to multiple age groups over time, with the ≤ 8, 9–13, and 14–18 year categories including the same cohort of 35 stallions at different life stages. In contrast, the ≥ 19 year category included fewer individuals (n = 28 stallions), as seven stallions had retired from breeding or died before reaching this age. Detailed counts for stallions and mares within each age category are provided in Supplementary Table S1.

Progeny racing performance data were obtained from Arion Pedigrees NZ (https://www.arion.co.nz/Home.aspx). For each stallion and foaling year, race records were reported as aggregate summaries of all progeny produced in that crop (e.g., total number of runners, winners, stakes performers, stakes winners and group winners etc.), rather than as individual horse-level performance histories, and the identity of individual foals within each summary was not available from this source. All summary reports were generated on the same date to ensure consistency with performance results. Recorded variables included the number of progeny that; participated in a race, won at least one race, placed in ‘stakes’ races, won at least one ‘stakes’ races, and won at least one ‘group’ race. All outcomes were treated as binary indicators at the progeny level within each foal crop summary. Stakes performance was defined as any first-, second-, or third-place finish in a stakes race, with all placings treated equivalently and no weighting applied to finishing position. Similarly, no weighting was applied for multiple stakes or group race wins, with each progeny counted only once per outcome category regardless of the number of stakes or group wins achieved. For clarity, a stakes race refers to a Listed or Group/Grade level race that carries higher prize money and prestige than standard races, while a group race is a subset of stakes races representing the most elite level of competition.

Statistical analysis

Breeding, fertility, and progeny performance data were collated in Microsoft Excel and outcomes such as seasonal fertility and performance percentages were calculated for each year of a stallion’s breeding life.

Statistical analyses were conducted in R (version 4.4.1) using generalized linear mixed-effects models (GLMMs) with a binomial error distribution and logit link function. Binary outcomes including number of foals, race runners, winners, stakes performers, stakes winners and group winners were modelled across four stallion paternal age categories: ≤ 8 years, 9–13 years, 14–18 years, and ≥ 19 years. For each of the above outcomes, sire age category was included as a fixed effect, and stallion identity was fitted as a random effect to account for repeated measures within stallions. Progeny year of birth, sire year of birth and covering year were also assessed as random effects to account for cohort- and time-related variation in environmental conditions. Model comparison using Akaike Information Criterion (AIC) indicated that only stallion identity and progeny year of birth improved the model fit and therefore were the only random effects retained in the final models. This modelling framework allowed estimation of sire age effects while accounting for both between-sire clustering and temporal cohort structure. Sire age was modelled as a categorical variable in the primary analyses to facilitate interpretation across biologically meaningful life-stage groups and align with industry-relevant classifications and was additionally explored as a continuous predictor in supplementary analyses (Supplementary Fig. S5). Proportion data were analysed using weights equal to the total number of eligible progeny within each category, corresponding to the binomial denominator. Estimated marginal means (EMMs) and corresponding 95% confidence intervals were calculated using the emmeans package³³. Odds ratios for pairwise comparisons among age categories were derived from the fitted models using the same package. Overdispersion was assessed using Pearson residuals as the ratio of the sum of squared residuals to the residual degrees of freedom, with no evidence of overdispersion detected.

Models were fitted using the lme4 package³⁴, and visualisations of model predictions and raw data were generated using ggplot2³⁵ and gridExtra³⁶. GLMM results are presented as estimated marginal means with 95% confidence intervals overlaid on raw data points.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualisation, C.J., R.A.G., A.S., N.A.H and Z.G.; Funding acquisition, A.S., R.A.G., and Z.G.; Supervision, A.S., R.A.G., N.A.H., and Z.G.; Methodology, C.J., R.A.G., and A.S.; Investigation, C.J.; Statistical analyses, C.J., and R.U. C.J wrote the paper with input and interpretation from all other authors.

Data availability

The authors declare that the aggregated data supporting the findings of this study are available within the paper and its supplementary information files. Any additional raw data are available on request from the corresponding author C.J.

Code availability

No custom code or algorithms were developed for this study. Statistical analyses were performed in R (version 4.4.1) using publicly available packages. Analysis scripts are available from the corresponding author C.J. upon reasonable request.

Competing interests

The authors declare no competing interests.