Camouflage mismatch and evolutionary rescue in snowshoe hares

Introduction

• Climate change imposes rapid, novel stressors; camouflage mismatch due to shorter snow duration is a critical example for seasonal color-molting species.
• Snowshoe hares are a model because background matching is essential for camouflage and predation risk is high.
• Research questions: how does camouflage mismatch affect fitness and can populations adapt plastically or evolutionarily?
• Key finding: mismatch costs reduce survival; there is substantial among-individual variation in molt timing that could enable evolutionary rescue.

Study design and area

• Location: two sites in Western Montana, Seeley Lake and Gardiner, ~330 km apart.
• Subjects: 186 radiocollared snowshoe hares monitored weekly for survival; coat color molt and snow cover observed weekly.
• Monitoring period: roughly 2009–2012; predators accounted for ~67% of deaths, with some right-censoring.
• Measurements:
  • Coat colour molt (percent white) estimated weekly (0–100%).
  • Snow cover within 10 m around each hare (0–100%).
  • Color contrast defined as the absolute difference between hare whiteness and mean snow cover for the site.
• Key threshold: camouflage mismatch defined as $ext{contrast}_{i,j} \geq 0.60$ (60%).

Measurements and modeling

• Color-contrast modeling: a nonlinear model to handle missing color observations; links whiteness to Julian day via individual random effects.
  • Form (illustrative):
  • W{i,j} = 100 ig(1 + \, \exp(-a{0,i} - a{1,i} \cdot JDj)\big)^{-1} + e
  • Where $W<em>{i,j}$ is whiteness, $JD</em>j$ is standardized Julian day, and $a<em>{0,i}, a</em>{1,i}$ are individual random effects.
• Color contrast used in survival analysis:
  • $ext{contrast}<em>{i,j} = |W</em>{i,j} - S<em>j|$, where $S</em>j$ is mean snow cover at the week/site.
• Survival modeling (Bayesian, known-fate):
  • Basic form: $\text{logit}(\pi<em>{i,j}) = b</em>0 + b<em>1 \cdot x</em>{i,j} + c_i$
  • $\pi<em>{i,j}$ = weekly survival probability, $x</em>{i,j}$ = covariate (color contrast), $c_i$ = individual random effect.
• Selection coefficient: temporally standardized weekly color contrast to estimate directional selection on mismatch.
  • Standardization: subtract weekly mean and divide by weekly SD; slope gives the standardized selection coefficient.
  • Result: standardized selection coefficient ≈ $-0.04$ (95% credible interval: $[-0.061, -0.017]$).
• Survival costs and population projections:
  • Weekly survival costs under mismatch used to project annual survival under future snow-duration scenarios.
  • Annual survival projection (two-weekend extremes):
  • $\text{AnnualSurv} = \mathcal{S}<em>{0\%\text{contrast}}^{c} \cdot \mathcal{S}</em>{60\%\text{contrast}}^{(52-c)}$
  • where $c$ = weeks with 0% contrast (matched weeks), and $52-c$ = weeks with 60% contrast.
• Future snow-duration scenarios:
  • Time periods: mid-century (2030–2059) and late-century (2070–2099).
  • Emission scenarios: IPCC AR5 RCP4.5 (medium-low) and RCP8.5 (high).
• Population dynamics:
  • Baseline hare population: asymptotic geometric growth rate $k = 1.15$ from a separate, intensive study.
  • Population-growth projections modify spring/fall survival according to mismatch costs; reproductive rates left unchanged.
  • Sensitivity analysis indicates post-weaning survival drives population dynamics in these models.

Key results

• Individual variation: strong among-individual variation in molt phenology, leading to notable variation in color contrast across individuals; some hares show >50% whiteness differences for ~7 weeks/year.
• Mismatch frequency: camouflage mismatch (contrast ≥ 60%) is relatively infrequent per individual ( < 1 week/year on average).
• Survival costs and selection:
  • Weekly survival cost of mismatch is negative and substantial:
  • $\mathcal{S}<em>{\text{match}} \approx 0.96\, ,\quad \mathcal{S}</em>{60\%} \approx 0.92\, ,\quad \mathcal{S}_{100\%} \approx 0.89$ per week (approximate values from the model).
  • Per-week effect size on survival for color contrast: $b_{\text{Contrast}} = -0.95$ (95% CRI $[-1.82, -0.035]$).
  • Under temporally standardized weekly contrast, standardized selection coefficient: $\approx -0.04$ (95% CRI $[-0.061, -0.017]$).
  • No detectable difference in mean survival between the two study sites: $b_{\text{Site}} = 0.004\, (95\%\ CRI: [-0.54, 0.52])$.
• Annual survival projections (cost of 60% contrast):
  • Mid-century and late-century survival under high emissions (RCP8.5):
  • $\text{AnnualSurv}<em>{\text{mid,8.5}} \approx 0.082\,,\; \text{AnnualSurv}</em>{\text{late,8.5}} \approx 0.070$
  • Baseline annual survival was $0.093$.
  • Under medium-low emissions (RCP4.5): $\text{AnnualSurv}<em>{\text{mid,4.5}} \approx 0.085\,,\; \text{AnnualSurv}</em>{\text{late,4.5}} \approx 0.080$
• Population growth rate (k) projections:
  • Baseline k = $1.15$.
  • Under RCP8.5: mid-century k ≈ $1.02$, late-century k ≈ $0.88$ (strong declines expected).
  • Under RCP4.5: mid-century k ≈ $1.05$, late-century k ≈ $1.00$ (near replacement by late century).
  • Overall, under high-emissions, demographic costs from mismatch could drive populations toward decline unless evolutionary rescue occurs.

Implications and conclusions

• Evolutionary rescue is potentially critical for maintaining camouflage-molting species under rapid snow-duration decline.
• Necessary conditions for rescue include: sufficient additive genetic variation in molt phenology, large population sizes, ongoing gene flow, and mitigation of anthropogenic stressors including climate change.
• Phenotypic plasticity could aid rapid responses, but its capacity to fully counteract increasing mismatch remains uncertain.
• The study provides direct field evidence of anthropogenic climate-change-induced selection on a highly variable trait and demonstrates how selection can translate into meaningful demographic consequences.

Key concepts to recall

• Camouflage mismatch: mismatch between an animal’s coat color and background due to changes in snow duration.
• Molt phenology: timing of seasonal coat-color changes (fall and spring molts).
• Evolutionary rescue: rapid evolutionary response that prevents population extinction under environmental change.
• Selection coefficient: standardized measure of how a trait affects fitness; here, about $-0.04$ for color contrast.
• Population growth rate: the long-term growth proxy $k$; values around 1 indicate replacement-level dynamics, >1 growth, <1 decline.
• Climate scenarios: RCP4.5 and RCP8.5 used to project future snow duration and mismatch frequency.

Equations snapshot (all in LaTeX)

• Color contrast definition:$ext{contrast}<em>{i,j} = |W</em>{i,j} - S_j|$
• Mismatch threshold:$ext{contrast}_{i,j} \geq 0.60$
• Survival model:$\text{logit}(\pi<em>{i,j}) = b</em>0 + b<em>1 \cdot x</em>{i,j} + c_i$
• Standardized contrast (selection):$z<em>{i,j} = \frac{\text{contrast}</em>{i,j} - \mu<em>{\text{week}}}{\sigma</em>{\text{week}}}$
• Annual survival under contrast levels:$\text{AnnualSurv} = \mathcal{S}<em>{0\%\text{contrast}}^{c} \cdot \mathcal{S}</em>{60\%\text{contrast}}^{52-c}$
• Baseline population growth:$k = 1.15$
• Projected k under scenarios (example values):
  • Middle of century (RCP8.5): $k \approx 1.02$
  • Late century (RCP8.5): $k \approx 0.88$
  • Middle of century (RCP4.5): $k \approx 1.05$
  • Late century (RCP4.5): $k \approx 1.00$