https://www.journals.uchicago.edu/doi/10.1086/742104
Although this is rotifers it applies to all sexual creatures (with mitochodria). I have explained what I think causes this and it is confusing because it involves probabilities.
chatGPT(5.6paid):
Paper
Liguori et al. (2026), “Transgenerational and Intergenerational Maternal Age Effects Exhibit Complex, Genotype-Specific Patterns of Inheritance,” The American Naturalist.
Summary
The paper asks whether the effects of maternal age on offspring fitness:
- accumulate over successive generations;
- disappear when maternal age is switched;
- follow the same pattern in different genotypes.
The authors studied two clonal strains of the rotifer Brachionus manjavacas: BmanL5 and BmanRUS. Because these rotifers reproduce parthenogenetically, genetic segregation and paternal effects are largely avoided.
They established lineages in which offspring were repeatedly collected from:
- young mothers, aged 3 days; or
- old mothers, aged 10–11 days.
These lineages were maintained for three offspring generations. In generation 3, they also switched maternal age, producing four histories:
- YYY: consistently young mothers;
- YYO: young lineage, then old mother;
- OOO: consistently old mothers;
- OOY: old lineage, then young mother.
Each offspring was followed throughout life. The measured outcomes included lifespan, lifetime reproductive output (LRO), maximum daily reproduction, reproductive timing, reproductive-period length and post-peak reproductive decline.
Main findings
1. No cumulative Lansing effect
Neither strain showed progressively worsening lifespan or reproduction over three generations of old mothers. This contrasts with Lansing’s classic rotifer experiments, in which repeated old maternal age supposedly transmitted a cumulative “aging factor.”
The absence of cumulative decline is the paper’s clearest result.
2. Maternal-age effects were strongly genotype dependent
The two strains sometimes responded in opposite directions:
- In BmanL5, offspring of old mothers had lower lifetime reproduction and, in the first two generations, lived about two days less than young-mother offspring.
- In BmanRUS, offspring of old mothers tended to live longer and had higher lifetime reproduction.
Thus, advanced maternal age was not intrinsically harmful. Its apparent fitness effect depended on genotype and trait.
3. Effects differed between traits
Even within a strain, an old mother could confer an advantage in one measure and a disadvantage in another. For example, offspring of old mothers generally had higher maximum daily reproduction, while sometimes having shorter reproductive periods or faster reproductive decline.
Maternal age therefore altered reproductive scheduling, not simply overall “quality.”
4. Lifespan effects weakened across generations
In BmanL5, the significant lifespan disadvantage associated with old mothers in G1 and G2 had disappeared by G3. BmanRUS showed a nonsignificant lifespan advantage from old mothers that also became smaller over successive generations.
This pattern is weakening rather than cumulative inheritance.
5. Switching maternal age produced incomplete and asymmetric reversibility
Reversibility depended on strain, trait and direction of switching:
- The negative effect of old maternal age on lifetime reproduction in BmanL5 was fully reversible.
- In BmanRUS, lifespan and reproductive-output effects were reversible only in the young-to-old direction.
- Some measures retained grandmaternal effects or showed maternal-by-grandmaternal interactions.
- Reproductive-senescence patterns were particularly complicated and were not simply reset by changing maternal age.
The inheritance pattern is consequently more accurately described as context-dependent interaction between recent maternal histories than as transmission of a single aging factor.
What is genuinely novel?
1. Factorial switching after several generations
The strongest novelty is the YYY/YYO/OOO/OOY design. Previous work has often compared young- and old-parent lineages, but this experiment switches maternal age after several generations and tests maternal and grandmaternal histories together.
This distinguishes:
- an effect caused predominantly by the current mother’s age;
- a persistent ancestral effect;
- interaction between current and ancestral maternal ages.
2. Direct demonstration of within-species genotype specificity
The paper shows that two genotypes of the same species can display opposite maternal-age effects under nominally identical conditions. That is more informative than merely showing differences between distantly related species.
It undermines the idea of a universal Lansing effect and suggests that maternal-age phenotypes can be genetically modulated.
3. Multiple dimensions of reproductive ageing
The authors do not restrict fitness to lifespan or total reproduction. They examine peak output, reproductive timing, reproductive-period duration and reproductive decline. This reveals that maternal age can redistribute reproduction across life without uniformly increasing or decreasing every fitness component.
4. Evidence against a simple cumulative inherited-damage model
The findings provide relatively strong phenotypic evidence that maternal-age effects in these strains are not explained by one progressively accumulating, uniformly detrimental factor. That is a useful empirical correction to the traditional Lansing-effect narrative.
Critique
Strengths
- The parthenogenetic rotifer model limits genetic segregation and paternal confounding.
- Individuals and maternal lineages were tracked across their complete lifespans.
- The switching design is conceptually strong.
- Two strains expose genotype dependence that a single-strain experiment would miss.
- The authors analyse several biologically distinct fitness measures.
- Data and analysis code are reported as publicly available.
Important limitations
1. Old-mother offspring come from a selected subset of mothers
To contribute to an old-mother cohort, a rotifer had to survive and remain reproductive until day 10 or 11. Young-mother offspring could be collected from almost every mother at day 3.
Consequently, “maternal age” is partly confounded with selection for mothers possessing:
- longer survival;
- later reproduction;
- better late-life condition;
- potentially distinctive inherited characteristics.
This is particularly important for the apparent benefits of old maternal age in BmanRUS. They could reflect selective reproduction by high-quality late-surviving mothers rather than a beneficial effect of ageing itself. Tracking matriline as a random effect does not fully remove this biological selection process.
A stronger design would compare early- and late-born offspring from the same mothers wherever possible, or explicitly model the probability of surviving and reproducing to the old collection age.
2. Generation is confounded with experimental time
Successive generations necessarily occurred at different calendar times. Changes between G1, G2 and G3 could therefore include:
- food-culture differences;
- incubator or handling drift;
- seasonal laboratory effects;
- experimenter or batch effects.
Generation-dependent weakening cannot be unequivocally attributed to transgenerational biology unless independent temporally staggered replicate lineages or batch controls were included.
3. Pre-experimental young-age conditioning biases the early generations
For two generations before G0, lineages were age-synchronised through young mothers. The authors acknowledge that the old line therefore began with a history of young mothers.
This means G1 and G2 are not equilibrium “old-line” generations. The apparent weakening of lifespan effects across generations could reflect the fading influence of pre-experimental young ancestry, rather than progressive attenuation of an old-mother signal. This is an especially important alternative interpretation because weakening across generations supports much of their epigenetic argument.
4. The mechanistic conclusions are speculative
The study measures phenotypes only. It does not measure:
- mitochondrial number, function or mtDNA heteroplasmy;
- mitophagy or mitochondrial stress responses;
- histone modifications;
- chromatin accessibility;
- small RNAs;
- transcriptomic changes;
- egg size or biochemical provisioning;
- oxidative damage or DNA mutations.
The conclusion that epigenetic or mitochondrial mechanisms are “more likely” is therefore a hypothesis generated by the inheritance pattern, not evidence that either mechanism operated.
5. Maternal provisioning is dismissed too readily
The authors argue that provisioning effects might be expected to accumulate across generations. That is not necessarily true. Egg provisioning is reconstructed by each mother and could respond immediately and reversibly to her current age, metabolic condition and ancestral phenotype.
Provisioning could therefore produce:
- noncumulative effects;
- one-generation reversibility;
- strain differences;
- maternal-by-grandmaternal interactions.
Without measuring egg size, lipid content, protein, metabolites or mitochondrial loading, provisioning remains a plausible explanation.
6. Mutation accumulation is not decisively excluded
A uniformly deleterious, rapidly accumulating mutation model is inconsistent with the results. But the experiment does not rule out all contributions from mutation or mtDNA variation.
Three generations may be too short to detect appreciable nuclear mutation accumulation. Furthermore, mitochondrial bottlenecks, intracellular selection and heteroplasmy segregation can generate nonlinear, reversible-looking or genotype-specific population phenotypes. The authors’ argument is therefore valid against a simple mutation model but broader than the data justify.
7. Only two genotypes were tested
The strains demonstrate that genotype specificity exists, but they cannot establish how common it is or explain its genetic basis. With only two strains, “genotype-specific patterns” could also include idiosyncratic strain histories or laboratory adaptation.
A larger strain panel and reciprocal environmental testing would be needed to connect maternal-age effects to ecological adaptation.
8. Many outcomes and contrasts increase interpretive flexibility
The paper examines numerous traits, generations, strains, interactions and post-hoc comparisons. This is biologically informative, but it also creates a substantial multiple-testing burden. Some reported patterns—especially asymmetric reversibility—depend on whether one contrast crosses the significance threshold while another does not.
Direct statistical comparison of effect sizes is preferable to inferring asymmetry from “significant in one direction, nonsignificant in the other.”
9. Some reproduction analyses may oversimplify senescence
Post-peak reproductive decline was modelled approximately linearly. Reproduction over age is commonly nonlinear and subject to floor effects and selective disappearance. Individuals with shorter lifespans contribute fewer late-life observations, so the estimated slopes can combine within-individual reproductive decline with changes in the composition of surviving individuals.
Overall assessment
This is a well-designed and useful phenotypic study whose principal contribution is to show that multigenerational maternal-age effects are neither universally detrimental nor predictably cumulative. They depend strongly on genotype, trait and ancestral sequence.
The experimental evidence supports rejection of a simple, universal Lansing-style accumulating aging factor. It does not, however, establish that histone modifications, small RNAs or mitochondrial quality-control mechanisms cause the observed inheritance. Maternal selection, provisioning and pre-experimental lineage history remain substantial alternative explanations.
The best interpretation is therefore:
The paper identifies complex, genotype-specific inheritance patterns and narrows the range of plausible mechanisms, but it does not yet identify the mechanism.
The most valuable follow-up would combine the same switching design with within-mother comparisons, egg-provisioning measurements, mitochondrial functional and heteroplasmy assays, and histone/small-RNA profiling.
further chatGPT: why do the papers authors disagree with the idea that this is inherited through the mtDNA germline
The authors do not completely reject mitochondrial inheritance. They reject progressive accumulation of permanent mtDNA mutations as the principal explanation.
Their reasoning is based on three observations.
-
The effects did not accumulate.
If old mothers progressively passed on additional harmful mtDNA mutations, the O → OO → OOO line should show worsening lifespan or reproduction. It did not. -
Some effects were reversible within one generation.
Changing from an old to a young mother sometimes reset the offspring phenotype. Permanent mtDNA sequence mutations would not ordinarily disappear merely because the next mother reproduced young. -
Old maternal age was sometimes beneficial.
In BmanRUS, old-mother offspring had greater lifetime reproduction and tended to live longer. The authors think random age-associated germline mutations should predominantly cause deterioration rather than opposite effects in different strains.
However, the paper explicitly leaves room for mitochondrial mechanisms other than stable mtDNA mutation inheritance. It suggests that old oocytes could transmit:
- oxidatively damaged mitochondrial proteins or membranes;
- altered mitochondrial abundance or physiological state;
- dysfunctional mitochondria that are subsequently removed by mitophagy;
- a mitochondrial stress signal that is reset through the mitochondrial unfolded-protein response.
These forms of mitochondrial inheritance could weaken or disappear in a young mother’s germline.
Why their argument is not decisive
Their reasoning only challenges a simple model:
old mother → more harmful mtDNA mutations → progressively more mutant mtDNA in each generation → steadily worsening offspring.
Real mtDNA inheritance can be much less linear because of:
- the mitochondrial genetic bottleneck;
- random heteroplasmy segregation;
- replicative selection between mtDNA variants;
- purifying selection during oogenesis;
- mitophagy selecting among mitochondria;
- threshold effects, where phenotype changes only above a heteroplasmy level;
- nuclear-genotype × mtDNA interactions.
These processes could generate noncumulative, strain-specific and apparently reversible population-level results. “Reversal” could mean that a young mother preferentially transmitted a lower-mutant mitochondrial subpopulation, not that an mtDNA mutation was repaired.
There is also an important distinction between mtDNA sequence inheritance and mitochondrial germline inheritance. An egg inherits whole mitochondria, including their:
- mtDNA heteroplasmy;
- proteins and lipids;
- membrane potential and structural state;
- metabolites;
- mitochondrial RNAs;
- associated quality-control signals.
Therefore, the observations are compatible with mitochondrial germline transmission broadly construed. They are simply less compatible with the authors’ relatively narrow model of steadily accumulating, irreversible mtDNA mutations.
Because they did not sequence mtDNA or measure heteroplasmy, copy number, respiration, membrane potential or mitophagy, the study cannot actually exclude mtDNA germline inheritance. A more accurate conclusion would be:
The results argue against uncomplicated cumulative inheritance of deleterious mtDNA mutations, but they do not distinguish epigenetic inheritance from dynamic mtDNA heteroplasmy or other inherited mitochondrial states.
further chatGPT: however, if there is a stochastic element to the mtDNA germline operating through an mtDNA bottleneck and process like follicular atresia then the results are as would be predicted by that mechanism is that not the case
Yes. If Brachionus has substantial pre-existing mtDNA heteroplasmy, a germline bottleneck followed by stochastic segregation and selective elimination of mitochondrial/oocyte lineages could produce patterns resembling those reported.
The paper therefore excludes only a simple cumulative-mutation model—not a dynamic mtDNA germline-selection model.
How that mechanism could generate the results
A plausible sequence would be:
- Mothers contain a mixture of mtDNA variants.
- A restricted mitochondrial sample populates each developing oocyte—the bottleneck.
- Random segregation produces oocytes with different heteroplasmy levels.
- Mitochondrial quality control eliminates or suppresses some germ cells, oocytes or mitochondrial populations.
- Maternal age changes the strength or direction of this selection.
- The surviving egg transmits a selected—but still stochastic—mtDNA population.
- The next generation repeats the bottleneck, potentially moving heteroplasmy back below a phenotypic threshold.
That could explain several findings.
Nonaccumulation
Repeated old maternal age would not necessarily cause progressively worse phenotypes. Purifying selection could remove the most dysfunctional mitochondrial populations, opposing new damage and drift.
Apparent reversibility
Switching to reproduction at a young age could produce eggs with lower deleterious heteroplasmy. The mutation has not literally been reversed; the sampled and transmitted mitochondrial population has changed.
Nonlinear grandmaternal effects
Successive bottlenecks can produce irregular trajectories:
[
h_{G1} \rightarrow h_{G2} \rightarrow h_{G3}
]
where (h) is heteroplasmy. There is no requirement that (h) move monotonically in one direction. Threshold effects could make modest heteroplasmy changes appear as disappearance or reappearance of a phenotype.
Genotype specificity
BmanL5 and BmanRUS could differ in:
- their starting mtDNA variants;
- bottleneck size;
- mtDNA replication bias;
- mitophagy efficiency;
- selection among germ cells or eggs;
- nuclear control of mitochondrial quality;
- compatibility between nuclear and mitochondrial genotypes.
The same mtDNA change could therefore be detrimental in one nuclear background but neutral or beneficial in another.
Trait specificity
Different traits may have different energetic or signalling thresholds. A heteroplasmy shift could improve early reproductive output while worsening reproductive duration or lifespan. That resembles the paper’s multidirectional results.
The follicular-atresia qualification
I would describe the rotifer mechanism as atresia-like selection, rather than necessarily follicular atresia itself. Brachionus does not possess mammalian ovarian follicles with the same granulosa-cell architecture and prolonged primordial-follicle reserve.
Nevertheless, functionally comparable selection could occur through:
- loss of germline nuclei;
- failure of oocytes to mature;
- selective mitochondrial destruction within oocytes;
- resorption or failure of developing eggs;
- differential reproductive success of oocytes;
- survival selection among mothers that remain reproductive at old age.
The last point is especially important in this experiment: only mothers that survived and continued producing offspring at 10–11 days could contribute to the old-mother cohorts. That creates an additional selection filter which could enrich particular nuclear–mitochondrial combinations.
What the hypothesis specifically predicts
A stochastic bottleneck should affect not only group means but also variance and lineage structure. One would predict:
- greater offspring-to-offspring heteroplasmy variance than mother-to-offspring mean change;
- clustering of phenotypes within matrilines;
- occasional extreme offspring rather than a completely uniform age effect;
- heteroplasmy shifts between early and late eggs from the same mother;
- associations between particular variants or total mutational burden and fitness;
- evidence of stronger purifying selection among late-produced eggs or old surviving mothers.
The paper mostly analyses mean lifespan and reproductive differences. It does not sequence mtDNA or test whether phenotypic variance expands and contracts across generations. It therefore does not test these distinctive bottleneck predictions.
Bottom line
Your proposed mechanism is not merely compatible with the results; it can naturally predict their stochasticity, nonaccumulation, apparent reversibility and genotype dependence. But those patterns are not unique to it: histone inheritance, small RNAs, provisioning and inherited mitochondrial damage could produce similar outcomes.
The authors are justified in rejecting a simple model of steadily accumulating irreversible mtDNA mutations. They are not justified in treating the experiment as evidence against a model involving:
mtDNA heteroplasmy → germline bottleneck → stochastic segregation → age-dependent purifying selection.
Distinguishing that model would require longitudinal single-mother sampling of early and late eggs, deep mtDNA sequencing, heteroplasmy-variance analysis and measurement of oocyte loss or mitochondrial selection.