This repository contains codes and data needed to reproduce the analyses and figures of the manuscript:

García-Gómez, G., Hirst, A.G., Spencer, M., & Atkinson, D. (2024). Responses of intraspecific metabolic scaling to temperature and activity differ between water- and air-breathing ectothermic vertebrates.

García-Gómez, G., Hirst, A.G., Spencer, M., & Atkinson, D. (2024). Data, code, and model outputs of the manuscript: Responses of intraspecific metabolic scaling to temperature and activity differ between water- and air-breathing ectothermic vertebrates. OSF (DOI 10.17605/OSF.IO/N6M5J): https://osf.io/n6m5j/

• Model_temperature_test.R This script contains all the code necessary to reproduce the results from the Bayesian model analysing the relationship between the scaling slope and the temperature-increased metabolic level within ectothermic vertebrate species.
• Model_activity_test.R This script contains all the code necessary to reproduce the results from the Bayesian model analysing the relationship between the scaling slope and the activity-increased metabolic level within ectothermic vertebrate species.

The latter scripts provide the option to run just the model with student-t (as in the present manuscript) or with Gaussian errors, as follows:

fitt <- # set TRUE to fit the model with student-t errors

fitgaussian <- # set TRUE to fit the model with Gaussian errors (pre-set as FALSE)

(see lines 49-50 in Model_temperature_test.R, and 41-42 in Model_activity_test.R)

Additionally, Model_temperature_test.R provides the option to (1) simulate 10 datasets under the temperature-effect model, with posterior mean parameter values and the same structure as the real data, and fitting the model to these simulated data sets; (2) plot estimates of the model fitted simulated data; and (3) check model residuals, as follows:

fittsimulated <- # set TRUE to fit the student-t model to data simulated under posterior mean parameters estimated from real data (pre-set as FALSE)

processtsimulated <- # set TRUE to plot estimates from student-t model fitted to simulated data (pre-set as FALSE)

tdiagnostics <- # set TRUE to get diagnostics for the model with student-t errors (pre-set as FALSE)

(see lines 51-53 in Model_temperature_test.R)

Last, the number of iterations in these models can be changed in line 180 of Model_temperature_test.R and line 169 of Model_activity_test.R.

There are two datasets used in the analysis, which are included in the appendix of manuscript as Table S1 and Table S2. These datasets contain data on intra-specific metabolic scaling of water- and air-breathing ectothermic vertebrates. Table S1 is used to test the effect of temperature, mediated by metabolic level, on the scaling slopes. Table S2 is used to test the effect of activity level, mediated by metabolic level, on the scaling slopes.

• Table S1, file: table_S1.csv contains data on metabolic scaling regressions performed at various temperatures from experiments in inactive animals. These experiments measured mostly resting or routine metabolism, which may include some spontaneous activity. The columns are:

  • group whether the species is water- or air-breather
  • species name of the species as reported in the original study
  • species_phylo name of the species as recorded in Open Tree of Life (https://tree.opentreeoflife.org)
  • experiment label for the experiment from which a set of scaling regressions was produced. Each experiment contains at least two regressions measured at two temperatures, whereas other conditions (e.g., metabolic state of animals) remain the same
  • temp experimental temperature (in degrees Celsius) at which the scaling regression was measured. If the originial study reported temperature ranges instead of a single temperature for each regression, the mean temperatures of the ranges were used
  • b metabolic scaling exponent or allometric slope (dimensionless) of a linear regression between log metabolic rate and log body mass
  • L metabolic level (in mg O₂ g^-1 h^-1), calculated as the mass-specific metabolic rate at the geometric mass-midpoint of the mass range of each regression
  • log10_L log₁₀-transformed metabolic level (in mg O₂ g^-1 h^-1)
  • mass_mid.g body mass (in g) at the geometric midpoint of the mass range of the scaling regression
  • min_mass.g minimum body mass (in g) of the mass range of the scaling regression
  • max_mass.g maximum body mass (in g) of the mass range of the scaling regression
  • L_state_study metabolic state of animals as reported in the original study ("NA" if this state is not reported)
  • study_id identification number of the original study from which the data were compiled (see word file with the reference list for Table S1 and S2)

• Table S2, file: table_S2.csv contains data on metabolic scaling regressions performed in animals at different activity levels, from experiments that sometimes cover various temperatures. These experiments measured from low activity levels as resting to routine metabolism (i.e., none or some spontaneous activity) to high activity levels as active and maximum metabolism (i.e., sustained activity or until exhaustion) by forcing locomotion. The columns are as in Table S1, plus two additional columns:

  • experiment2 label for a set of scaling regressions measured at a single temperature in the same experiment, containing at least two regressions measured at different activity levels, whereas other experimental conditions remain the same
  • comp_L label for the minimum and maximum activity levels that were measured in an experiment (e.g., "rest_max" means a experiment included measurements on resting and maximum metabolism)
  • L_state categories “minimal” or “maximal” activity assigned following studies’ description of metabolic measurements, where minimal activity refers to resting individuals, and maximal activity refers to maximum swimming or running speeds, or exhaustion

Some of the compiled studies here are present in both Table S1 and S2, since these studies contained metabolic scaling regressions of animals at various temperatures as well as activity levels (e.g., Du Preez et al. 1988; Wright 1986; Hölker 2003; Gifford et al. 2013; see list of references), measuring from resting or routine (Table S1) to active states (Table S2).

All data processing was carried out in the R software version 4.2.0. The R folder contains the scripts to reproduce the statistical analyses and figures presented in this manuscript. To improve clarity, figures were slightly edited by the addition of side annotations and shapes.

The R packages used for each R script are enlisted in the corresponding R session files. Note that a C++ compiler is required for running these scripts.

This repository was provided by the authors under the MIT License.

In case of further questions, please contact: Guillermo García-Gómez, email: [email protected]