Research ArticleCLIMATOLOGY

Climatic warming destabilizes forest ant communities

See allHide authors and affiliations

Science Advances  26 Oct 2016:
Vol. 2, no. 10, e1600842
DOI: 10.1126/sciadv.1600842
  • Fig. 1 Experimental chambers warm the forest floor inhabited by ants in nest boxes.

    (A) Geographic position of warming arrays at Duke Forest (orange) and Harvard Forest (green) toward the center and northern boundary of temperate deciduous forest (45) and the local spatial arrangement of the chambers at each site. Color intensity indicates greater MAT, with chamberless control plots indicated by unshaded symbols. (B) A single warming chamber at Duke Forest. Note that the diameter of each chamber is roughly 1000 ant body lengths. (C) Nest box containing a Crematogaster lineolata colony, with the cover tile removed.

  • Fig. 2 Direct and indirect effects of warming on ant communities.

    Path diagram indicating the magnitude (arrow width, scaled for each model term separately; the one interaction term was arbitrarily given an intermediate arrow width) and direction (blue, negative; red, positive) of the effects of MAT and the presence of other ant species on the occupancy, colonization, and extinction of ant species inhabiting artificial nest boxes at the Duke Forest and Harvard Forest warming arrays (statistical summaries are given in table S2). Interaction effects between MAT and species presence are indicated by lines connecting MAT with species and terminating in a single arrow. Only statistically significant (P < 0.05) effects are shown. Species codes: As, Aphaenogaster spp.; Bc, B. chinensis; Cl, C. lineolata; Cs, Camponotus spp.; Ls, Lasius spp.; Ms, Myrmica spp.; Tc, T. curvispinosus; Tl, T. longispinosus.

  • Fig. 3 Stability and demographic responses of ant communities to warming.

    (A) Damping ratio, (B) occupancy, (C) colonization, and (D) extinction as functions of MAT (°C) for ant communities inhabiting nest boxes at Duke Forest (orange) and Harvard Forest (green); chambered plots are represented by filled symbols, and chamberless control plots are represented by open symbols. For the damping ratio, dashed lines represent simple linear regressions; the solid lines are from an analysis of covariance (ANCOVA) with separate intercepts for site and a common slope for MAT (Table 1). For occupancy, colonization, and extinction, mean proportions and binomial 95% confidence intervals are presented; dashed lines are predicted values from quasi-binomial GLMs (Table 1). Inset panels depict the null expectations for stability, occupancy, colonization, and extinction under a simple model of increasing activity of thermophilic ectotherms at higher temperatures.

  • Table 1 Occupancy, colonization, extinction, and stability responses to warming.

    For occupancy, colonization, and extinction, slope estimates ± 1 SE are from GLMs using a quasi-binomial error structure to examine the effect of MAT on community-wide occupancy, colonization, and extinction. F ratios and P values indicate the statistical significance of chamber temperature or site. The pseudo-r2 is calculated as 1 − (residual deviance / null deviance). For stability, slope estimates ± 1 SE are from simple linear regressions to examine the effect of MAT on stability (damping ratio) for each site considered separately and from an ANCOVA with separate intercepts for site and a common slope for MAT (the site × MAT interaction was not significant and was dropped from the final model).

    Data subsetResponsePredictorEstimateSEFPr2
    Duke ForestOccupancyMAT0.09650.029910.50.001560.0721
    ColonizationMAT0.03380.04220.640.4250.00552
    ExtinctionMAT−0.08340.037.750.006340.0678
    Harvard ForestOccupancyMAT0.03620.03561.030.3130.00815
    ColonizationMAT0.006260.03750.02780.8680.000221
    ExtinctionMAT−0.06270.036430.08680.026
    Duke ForestStabilityMAT−0.160.07934.0570.06510.238
    Harvard ForestStabilityMAT−0.250.09217.380.01870.381
    Both sites (common slope model)StabilityMAT−0.2010.059811.30.002370.347
    Site−0.8610.4184.250.0494

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/2/10/e1600842/DC1

    Levins’ metapopulation, turnover, and persistence models

    Alternative demographic and transition matrix model specifications

    Linking altered community dynamics with changes in community composition

    fig. S1. Frequency of nest box censuses at the two experimental warming arrays.

    fig. S2. Mean proportion of nest boxes occupied per chamber at Duke Forest.

    fig. S3. Mean proportion of nest boxes colonized per chamber at Duke Forest.

    fig. S4. Mean proportion of nest boxes that went extinct per chamber at Duke Forest.

    fig. S5. Mean proportion of nest boxes occupied per chamber at Harvard Forest.

    fig. S6. Mean proportion of nest boxes colonized per chamber at Harvard Forest.

    fig. S7. Mean proportion of nest boxes that went extinct per chamber at Harvard Forest.

    fig. S8. Equilibrium frequencies as a function of chamber temperature for each of the four focal species and empty nest boxes at Duke Forest.

    fig. S9. Equilibrium frequencies as a function of chamber temperature for each of the five focal species and empty nest boxes at Harvard Forest.

    fig. S10. Mean proportion of nest boxes occupied at equilibrium using Levins’ colonization-extinction formula at Duke Forest.

    fig. S11. Mean proportion of nest boxes that turn over at Duke Forest.

    fig. S12. Mean proportion of nest boxes occupied at equilibrium using Levins’ colonization-extinction formula at Harvard Forest.

    fig. S13. Mean proportion of nest boxes that turn over at Harvard Forest.

    fig. S14. Mean proportion of nest boxes that persisted to the next census per chamber at Duke Forest.

    fig. S15. Mean proportion of nest boxes that persisted to the next census per chamber at Harvard Forest.

    table S1. Observed transitions in the nest boxes at Duke and Harvard forests.

    table S2. Models of species associations at Duke and Harvard forests.

    table S3. Models of temperature effects at Duke Forest.

    table S4. Models of temperature effects at Harvard Forest.

    table S5. Models of community-wide responses.

    table S6. Models of equilibrium frequency as functions of temperature.

    table S7. Temperature dependence of individual transition probabilities.

    table S8. Transition matrix correlates of community stability.

  • Supplementary Materials

    This PDF file includes:

    • Levins’ metapopulation, turnover, and persistence models
    • Alternative demographic and transition matrix model specifications
    • Linking altered community dynamics with changes in community composition
    • fig. S1. Frequency of nest box censuses at the two experimental warming arrays.
    • fig. S2. Mean proportion of nest boxes occupied per chamber at Duke Forest.
    • fig. S3. Mean proportion of nest boxes colonized per chamber at Duke Forest.
    • fig. S4. Mean proportion of nest boxes that went extinct per chamber at Duke Forest.
    • fig. S5. Mean proportion of nest boxes occupied per chamber at Harvard Forest.
    • fig. S6. Mean proportion of nest boxes colonized per chamber at Harvard Forest.
    • fig. S7. Mean proportion of nest boxes that went extinct per chamber at Harvard Forest.
    • fig. S8. Equilibrium frequencies as a function of chamber temperature for each of the four focal species and empty nest boxes at Duke Forest.
    • fig. S9. Equilibrium frequencies as a function of chamber temperature for each of the five focal species and empty nest boxes at Harvard Forest.
    • fig. S10. Mean proportion of nest boxes occupied at equilibrium using Levins’ colonization-extinction formula at Duke Forest.
    • fig. S11. Mean proportion of nest boxes that turn over at Duke Forest.
    • fig. S12. Mean proportion of nest boxes occupied at equilibrium using Levins’ colonization-extinction formula at Harvard Forest.
    • fig. S13. Mean proportion of nest boxes that turn over at Harvard Forest.
    • fig. S14. Mean proportion of nest boxes that persisted to the next census per chamber at Duke Forest.
    • fig. S15. Mean proportion of nest boxes that persisted to the next census per chamber at Harvard Forest.
    • table S1. Observed transitions in the nest boxes at Duke and Harvard forests.
    • table S2. Models of species associations at Duke and Harvard forests.
    • table S3. Models of temperature effects at Duke Forest.
    • table S4. Models of temperature effects at Harvard Forest.
    • table S5. Models of community-wide responses.
    • table S6. Models of equilibrium frequency as functions of temperature.
    • table S7. Temperature dependence of individual transition probabilities.
    • table S8. Transition matrix correlates of community stability.

    Download PDF

    Files in this Data Supplement: