BGPT: Paper Review: The effect of dominance rank on female reproductive success in social mammals

Explore by Goal

Quick Explanation Copied

What this meta-analysis suggests

Across 444 effect sizes from 187 studies on 86 mammal species, higher female dominance rank is generally associated with higher reproductive success (overall positive effect), with the relationship stronger for reproduction than survival and strongly modulated by social environment (especially cooperative breeding) rather than by ecological harshness.

Evidence base:

Long Explanation

Paper Review (rigorous, skeptical, evidence-based)

Target paper: The effect of dominance rank on female reproductive success in social mammals ()

1) Visual synthesis of the paper’s headline quantitative claims

Note: the graphs below use only numeric values explicitly stated in the paper text you provided (not reconstructed from missing raw effect-size data).

Fig A. Reported effect-size magnitudes by reproductive trait class (direction: higher rank → higher outcome)

Shown values are the paper’s upper bounds for the main frequentist/bayesian reported intervals for each trait class (as quoted in your text). The underlying paper reports strongest effects for lifetime reproductive success and infant production, with the weakest for infant survival.

Fig B. Breeding system: cooperative breeders show higher dominance effects than plural/associated breeders

The paper reports cooperative breeders having substantially larger effects (mean ≈0.58) than plural (≈0.25) or associated breeders (≈0.23).

Fig C. Social context: larger group size and philopatry vs dispersal patterns in the paper

Paper statements used here: (i) effects were higher when females disperse to breed (mean ≈0.46) vs philopatric (mean ≈0.26); (ii) dominance effects increase with population density (paper reports positive association but your excerpt does not provide a single mean value for density).

2) Methodological audit (what the paper did, and why it matters)

2.1 Effect sizes & comparability across studies

The authors convert reported statistics into Z-transformed correlation coefficients (Zr), unifying heterogeneous study designs into a common metric, and explicitly align the sign so “positive” reflects higher-ranking females having higher reproductive output/survival/shorter interbirth intervals depending on the outcome definition.

They also correct a small number of entries for an earlier effect-size/variance formula mismatch, explicitly documenting the issue (effectRefs 425, 427, 428) and correction before analyses.

2.2 Bias diagnostics & skepticism about the literature

The authors report initial funnel-plot asymmetry and interpret it as a combination of publication bias, study-system bias, and study-time bias; they then model how known predictors shift the funnel-plot imbalance.

This is important because effect-magnitude estimates can be biased when only easy-to-detect systems (e.g., cooperative breeders with strong skew) are over-sampled. The paper directly acknowledges this mechanism through “study system bias” and its partial interaction with sample-size/time.

2.3 Statistical modeling: phylogeny-aware multilevel meta-analysis

The paper uses (i) frequentist multilevel meta-analytic models via metafor and (ii) Bayesian hierarchical models via rethinking/Stan, treating sampling variance as measurement error and modeling shared phylogenetic history via a covariance structure from a consensus MCC phylogeny.

3) Key biological conclusions, with explicit uncertainty boundaries

3.1 Core result: dominance rank generally tracks reproductive success

The authors report consistent evidence that higher-ranking females show higher reproductive success overall, even while acknowledging sample bias.

Confidence boundary: the magnitude is an association between rank and measured fitness proxies across within-group observations; it does not by itself prove causality or rule out trait confounding (e.g., condition/ability influencing both dominance and reproduction). The paper itself raises “phenotypic selection vs chance vs shared traits” as unresolved pathways.

3.2 Life-history moderator: rank benefits are stronger for reproductive output than survival

The paper finds the rank→fitness association is strongest for reproductive-output traits (e.g., lifetime reproductive success, age at first conception, infant production and interbirth interval) and weakest for infant survival.

3.3 Social environment matters more than ecological harshness (in their dataset)

Contrary to their harshness-based prediction, they report no evidence that environmental harshness increases the dominance advantage; instead, social context consistently mitigates or amplifies rank differences (notably cooperative breeding; and effects vary with group properties such as size and female dispersal).

4) Critical blind spots & ways this could mislead

4.1 The “rank” construct is heterogeneous

Dominance rank is operationalized differently across studies (agonistic interactions vs correlates; continuous vs categorical; different rank assignment conventions). The paper shows that effect-size estimates depend on these measurement choices, including categorical vs continuous rank scoring and how rank is inferred.

This creates an inference risk: part of what looks like “social moderation” could be statistical coupling between measurement practices and specific social systems (e.g., cooperative breeders often yield a single dominant vs a class of subordinates).

4.2 Association ≠ causation; trait confounding remains plausible

The paper’s own discussion emphasizes unresolved pathways: dominance could be a cause, or it could co-vary with intrinsic condition/quality, social opportunity, or chance in extreme conditions.

5) What would disprove or sharply revise the paper’s headline pattern?

5.1 Test disproof logic (falsification targets)

Null dominance effect after controlling for measurement and study-system: if new, preregistered sampling across underrepresented systems (where dominance–fitness contrasts are hard to detect) yields no positive mean association.
Reversal in causal estimates: if within-population causal designs (not just observational correlates) show dominants do not increase reproductive output beyond what would be predicted from condition/opportunity.
Measurement invariance failure: if improved harmonization of rank operationalization eliminates what now appears as rank-related reproductive skew.

These disproof directions follow directly from the paper’s own acknowledged limitations: bias sources, heterogeneity in how rank is measured, and inability to infer causality unambiguously.

6) Bespoke follow-up on BGPT (science actions)

Author reviews on BGPT

Feedback:

Updated: April 17, 2026