Body size predicts how long ant workers live - but not how they age or how they die from heat

Alana Moscardi; Rafael da Silva; Gleycon Silva

Source-linked AI summary

Body size predicts how long ant workers live - but not how they age or how they die from heat

Alana Moscardi, Rafael da Silva, Gleycon Silva

arXiv:2608.14245v1Aug 14, 2026q-bio.PEcs.LG

TL;DR

Mortality risk in social insects may combine dimensions with different predictors, but their separation has been limited. This study tests size-related longevity alongside senescence and thermal vulnerability, finding that body size predicts duration while circadian regime and temperature patterns govern other dimensions.

Problem
Body size predicts worker longevity, but its explanatory scope across senescence and thermal vulnerability remains limited.
Method
The study separates mortality risk into nested levels and tests whether size-related longevity is modulated by distinct mechanisms.
Results
Body size predicts longevity duration independently of colony size and temperature; senescence tracks circadian regime, and thermal risk plateaus above 20 °C.
Takeaways & Limitations
Mortality duration, senescence, and thermal vulnerability are decoupled and associated with different predictors, making body size an incomplete descriptor of mortality risk.
Takeaways & Limitations
With N = 18 species, circadian niche and subfamily identity cannot be fully separated, making their attribution operationally indistinguishable.

Abstract

from arXiv · show

In social insects, mortality risk comprises distinct components that may not share the same predictors: lifespan duration, senescence trajectory, and thermal vulnerability. We tested these three axes in 18 Australian ant species using paired field-laboratory survival assays (2,363 cohort-day observations; 1,148 workers). Body size predicted duration (Cox HR = 0.67, p = 0.002), while colony size (p = 0.60) and the size x temperature interaction (p = 0.72) showed no detectable moderating effect. A weak but significant size x foraging-rate interaction was detected (LRT p = 0.014), suggesting that intrinsic physiology remains the most parsimonious explanation for the main size-longevity pattern, although ecological context may contribute. Senescence trajectory was associated with circadian niche rather than size: it was steepest in matinal species (Kruskal-Wallis p = 0.009; matinal vs. crepuscular p = 0.002) and was uncorrelated with body mass (Spearman p = 0.32). Thermal hazard plateaued above 20 degrees C (Delta AIC = -38; p < 0.001), with elevated thermal sensitivity in Rhytidoponera (Ectatomminae) above the plateau (5% per degree C, p = 0.015). Circadian regime and lineage identity, not body size, therefore emerge as the most climate-relevant axes, although they are strongly collinear (Cramer's V = 0.85). These results show that body size captures only one dimension of mortality risk and that size-based vulnerability indices may misrank taxa when senescence and thermal sensitivity are decoupled from body size.

1 | INTRODUCTION

The introduction argues that ant mortality risk comprises lifespan, senescence trajectory, and thermal vulnerability, which may have distinct predictors. It frames three questions testing whether body size predicts all three axes or only longevity.

Conceptual framework: Mortality risk includes average lifespan, age-related mortality trajectory, and vulnerability to environmental heat, so body size may not predict all components.The introduction warns that a size-centric framework could mispredict which workers and lineages are most at risk under environmental change.
Study contribution: The unified paired field–laboratory framework tests the three mortality components together and finds no shared predictor detectable at N = 18 species.The study extends prior work on the same assemblage by partitioning mortality risk into longevity, senescence, and thermal vulnerability.
Q1 — Mechanism linking body size to longevity: Q1 tests whether the size–longevity association reflects intrinsic physiology, environmental buffering, or colony-level dilution, including moderation by colony size, foraging context, or temperature.These hypotheses distinguish lower mass-specific metabolism and oxidative stress from buffering against desiccation or heat and from colony-level effects.
Q2 — Mortality trajectory: Q2 asks whether senescence trajectory tracks body size or circadian activity regime, treating trajectory as potentially independent of average longevity.The Weibull shape parameter ρ distinguishes mortality acceleration with age (ρ > 1) from deceleration (ρ < 1).
Q3 — Temperature and taxon-specific thermal risk: Q3 asks whether thermal mortality scales linearly with temperature and whether vulnerability differs among ant lineages, especially at the subfamily level.The question is motivated by increasing temperatures and heat events and by the limited subfamily-level quantification of thermal vulnerability.

2 | MATERIAL AND METHODS

The study evaluates three mortality dimensions—duration, senescence trajectory, and thermal vulnerability—using distinct inferential units and survival models. Analyses combine cohort-day, colony, and species-level data while recognizing limited power for species-level and higher-taxonomic comparisons.

Analytical roadmap: The three claims test size-linked duration, circadian associations with senescence trajectory, and nonlinear, lineage-patterned thermal hazard through falsifiable sub-tests H1–H9.The strategy uses mechanistic exclusion, moderator tests, temperature nonlinearity, and subfamily effect heterogeneity rather than bivariate association alone.
Data and inferential levels: 2,363 cohort-day observations from 106 cohorts and 39 colonies formed the primary survival-analysis dataset, including 1,826 recorded daily mortality events.Cohort-day Cox models, temperature splines, and subfamily × temperature interactions were analyzed at this level.
Analytical methods: Species-level analyses used Weibull ρ, Kruskal–Wallis tests, Spearman correlations, PGLS, and variance partitioning across 18 species or filtered subsets.Species-level tests were treated as less powerful than cohort-day Cox models.
Data and inferential levels: 1,148 individual workers from 39 colonies and 18 species comprised the paired laboratory dataset under controlled temperature and humidity.The field dataset covered 18 Australian ant species monitored at the La Trobe Wildlife Sanctuary.
Limitations: N = 18 species was sufficient to estimate Cox and Weibull parameters but insufficient for fully factorial inference across all three mortality axes.Genus- or subfamily-level contrasts based on few tip taxa were interpreted provisionally.

| Replication Statement

The replication framework maps each claim to a single pre-specified primary endpoint and uses complementary controls for autocorrelation, phylogenetic structure, and robustness. The three endpoints are duration, senescence trajectory, and thermal plateau.

Primary endpoints: Each claim maps to one pre-specified primary endpoint: Cox M0 body-size hazard ratio for duration, laboratory Weibull shape ρlab for senescence, or AIC-selected piecewise Cox thermal plateau.The senescence endpoint is tested across circadian niches by Kruskal–Wallis, while breakpoint significance uses a Davies permutation test.
Confounding control: Cluster-robust standard errors, with cluster = cohort, absorb within-cohort autocorrelation in the survival analyses.The event is the daily mortality indicator, and time is measured as days since cohort formation.
Confounding control: Subfamily fixed effects absorb shared phylogenetic baseline hazard, with Dolichoderinae as the reference category.Sensitivity analyses use species fixed effects, stratified Cox models, and exclusion of pooled-Weibull fallback species.
Inference framework: The three claims address orthogonal null hypotheses by design, so no multiple-comparison adjustment is applied.This decision is stated as part of the replication framework rather than as an outcome of the analyses.

3 | RESULTS

Across 18 Australian ant species, body size predicted lifespan duration but not senescence trajectory or thermal sensitivity. Senescence varied with circadian niche, while thermal hazard plateaued above 20 °C and was elevated in Rhytidoponera.

Body size and longevity: HR = 0.67: each log10 unit increase in body mass reduced daily hazard by 33% (95% CI [0.52–0.87], p = 0.002).The size–longevity relationship was a between-species comparative pattern and showed negligible phylogenetic structure.
Contextual moderation: p = 0.60: colony size did not moderate the body-size effect, whereas foraging rate showed a weak but significant interaction (LRT χ2(1) = 6.08, p = 0.014).The temperature interaction was also null (HR = 1.00, 95% CI [0.98–1.03], p = 0.72), supporting intrinsic physiology as the parsimonious account while warranting caution about the foraging signal.
Senescence trajectory: 14 of 18 species displayed senescent laboratory mortality trajectories, but senescence was uncorrelated with body mass (Spearman ρ = −0.25, p = 0.32).Senescence differed among circadian niches (Kruskal–Wallis H = 9.35, p = 0.009), with matinal species steeper than crepuscular species (p = 0.002).
Thermal hazard: ΔAIC = −38: thermal hazard was sharply nonlinear, rising below 20 °C (HR = 1.216 per °C, p < 0.001) but nearing a plateau above 20 °C (HR = 1.016 per °C, p = 0.006).The breakpoint was T* = 20 °C air temperature, with 65.7% of cohort-day observations above the threshold.
Thermal hazard: 5% per °C: Rhytidoponera showed elevated thermal sensitivity above the assemblage response (HR = 1.05, 95% CI [1.01–1.09], p = 0.015).During extreme heat events above 30 °C, Rhytidoponera experienced 66% excess hazard (HR = 1.66, 95% CI [1.11–2.48], p = 0.014).

4 | DISCUSSION

The discussion reframes worker mortality risk as three partly decoupled axes: duration, senescence trajectory, and thermal vulnerability. Body size predicts duration, whereas circadian regime and lineage identity structure the other axes, limiting size-only vulnerability forecasts.

Three-axis reframing: No single predictor accounts for all three mortality components: body size explains duration but is decoupled from senescence trajectory and thermal vulnerability.The paper’s primary conceptual advance is replacing a size-centric framework with a three-axis framework for predicting environmental responses.
Senescence trajectory: Spearman ρ = −0.25 and p = 0.32 show no detectable size–trajectory association, while trajectory shape varies with circadian niche and is steepest in matinal species.The discussion describes the pattern as a graded matinal-to-crepuscular continuum rather than a sharply separated group effect.
Thermal vulnerability: ∆AIC = −38 and Davies p < 0.001 support a thermal-hazard plateau at T ∗= 20 °C, whereas Rhytidoponera shows excess hazard 5 % per °C.Extreme heat above 30 °C is associated with 66 % excess mortality in Rhytidoponera, but lineage-level thermal conclusions remain provisional.
Forecasting implications: Size-only vulnerability frameworks capture duration but can miss climate exposure structured by circadian regime and lineage identity, which are strongly collinear in this assemblage.Forecasting should emphasize extreme-heat frequency rather than only mean warming because behavioural buffering fails above 30 °C.
Duration: HR = 0.67 and p = 0.002 support a body-size effect on duration, while the size × temperature interaction CI [0.98, 1.03] provides no evidence of thermal moderation.The size effect is interpreted as a species-level comparative signal, not an individual-level effect; the foraging-rate interaction is a modest contextual qualification.

5 | CONCLUSION

Ant worker mortality is organized along three axes with different predictors: body size predicts longevity, circadian regime predicts senescence, and thermal risk plateaus above 20 °C. These findings show that size-based vulnerability rankings can misidentify taxa and ecosystem functions at risk from warming.

Body size predicts longevity duration independently of colony size and temperature, with only a weak qualification involving foraging rate.This supports intrinsic physiology as the main explanation for the size–longevity pattern.
Senescence trajectory is associated with circadian activity regime, not body size.
Thermal risk is nonlinear, plateauing above 20 °C, with elevated sensitivity in the Rhytidoponera lineage (Ectatomminae; Pagel’s λ = 0).The elevated sensitivity persists above the plateau and intensifies during extreme heat.
The trait predicting how long a worker lives does not predict how it ages or dies from heat.
Predicting ant responses to warming requires tracking extreme-heat frequency and weighting circadian regime and lineage identity rather than body size.These factors should guide rankings of taxa and ecosystem functions most at risk.

Data and Code Availability

Raw data and analysis code are publicly available through Dryad and a GitHub repository. A frozen Zenodo archive and README support reproducible execution and output generation.

Raw data are available via Dryad under Riskas et al. (2026), doi: 10.5061/dryad.j0zpc86s4.
Analysis code is available as a fully reproducible Python pipeline in source_files/ at https://github.com/[FILL-BEFORE-SUBMISSION].
A frozen archive is deposited on Zenodo, and the repository README documents execution order and expected outputs.Zenodo DOI: https://doi.org/[FILL-BEFORE-SUBMISSION].

Supplementary Tables

Supplementary analyses support the main body-size longevity signal while identifying limits to its interpretation. They find no proportional-hazards violation, no detectable phylogenetic signal, and strong collinearity between circadian niche and subfamily.

Proportional hazards: All p ≥0.17, so the proportional-hazards assumption was not violated for any primary covariate.The test used Schoenfeld residuals for the main Cox model with 2,363 cohort-day observations and 1,826 events.
Species-frailty robustness: 0.672 was the body-size hazard ratio in the reference model, while species-stratified Cox gave HR = 0.710, p = 0.354.The hazard-ratio direction remained below 1 across models, consistent with larger workers living longer, but species-controlled variants were non-significant.
Phylogenetic robustness: Pagel’s ˆλ = 0.00 (LRT χ2 (1) = 0.00, p = 1.00) indicated no detectable phylogenetic signal in subfamily thermal-sensitivity comparisons.Four of 18 species were excluded because species-specific Cox models failed to converge, and the Rhytidoponera/Ectatomminae effect was directionally consistent but non-significant in PGLS.
Phylogenetic robustness: 0.06 0.90 was the intercept-only PGLS body-size result, supporting retention of the main Cox analysis without phylogenetic correction.The analysis included 14 converged species; four were excluded because of convergence failure.
Variance partitioning: Cramér’s V = 0.85 accompanied adjusted R 2 values of 0.33 for niche-only and 0.19 for the full niche-plus-subfamily model.The reduced combined fit reflects strong collinearity rather than additive contributions from both predictors.