fit_all: Get global effect size and its standard error as well as...

In radchukv/adRes: Assessing Whether Phenotypic Changes in Response to Climate are Adaptive

Description

fit_all Subselects the data according to the specified requirements, extracts the effect sizes per each study for a specied condition and fits a meta-analytical model to calculate the global effect size and its standard error

Usage

fit_all(data, temperature = FALSE, precipitation = FALSE,
  phenology = FALSE, morphology = FALSE, condition, nb_cores,
  rand_trait = FALSE, fixed = NULL, digit = 3)

Arguments

data	A dataframe containing per each study the time series of relevant variables (i.e. yearly climate values, yearly trait values or yearly selection differentials) to be analyzed.
temperature	A boolean indicating if the temperature data should be extracted (default is FALSE).
precipitation	A boolean indicating if the precipitation data should be extracted (default is FALSE).
phenology	A boolean indicating if the phenological data should be extracted (default is FALSE).
morphology	A boolean indicating if the morphological data should be extracted (default is FALSE).
condition	A character specifying which condition is to be tested (for more details see Radchuk et al. (in review)): '1' - Condition 1 (effect of year on climate); '2' - Condition 2 (effect of climate on traits); '3' - Condition 3 (testing whether weighted mean selection over years differs from 0); '2b' - Condition 2, but also with abundance included as a predictor; '3b' - testing whether selection changes over years.
nb_cores	An integer indicating how many cores should be used for assessing the LRT by bootstrap (see spaMM documentation for more details).
rand_trait	A boolean indicating whether to include a trait type as a random effect in the model.
fixed	A string indicating the name of the fixed effect (called exactly as it is called in the dataframe meta_data) if a meta-analytic model includes a fixed predictor, and NULL (default) otherwise.
digit	An integer indicating how many digits to display on the screen.

Details

The data can be subselected per climatic variable and per trait category, as detailed in prepare_data. Next, a condition tested should be specified as described in extract_effects_all_ids. And finally, a meta-analytical model may be fitted with an intercept only to assess the global effect size across the studies or with a specified fixed effect.

Value

A list with three elements. The first element is the data that was subset according to the specified requirements. The second element is a dataset of extracted effects sizes per study and their standard errors, together with relevant meta-data per study. The third element is the list of length eight, containing the results of the meta-analysis: a global slope, its SE, a full model fitted with REML, a null model fitted with ML, a global model fitted with ML, results of LRT comparing the model with the effect vs. the reduced one, the data used for the meta-analysis, and heterogeneity metrics.

Examples

nb_cores <- 2L ## increase the number for using more cores
meta_Sel_phen <- fit_all(data = dat_Sel, temperature = TRUE,
                         precipitation = FALSE,
                         phenology = TRUE, morphology = FALSE,
                         condition = '3', nb_cores = nb_cores,
                         rand_trait = FALSE, fixed = NULL, digit = 3)

meta_Sel_morph <- fit_all(data = dat_Sel,
                         temperature = TRUE, precipitation = FALSE,
                         phenology = FALSE, morphology = TRUE,
                         condition = '3', nb_cores = nb_cores,
                         rand_trait = FALSE, fixed = NULL, digit = 3)

meta_Sel_phen_Fitn <- fit_all(data = dat_Sel,
                              temperature = TRUE, precipitation = FALSE,
                              phenology = TRUE, morphology = FALSE,
                              condition = '3', nb_cores = nb_cores,
                              rand_trait = FALSE,
                              fixed = 'Fitness_Categ', digit = 3)

radchukv/adRes documentation built on June 1, 2019, 7:05 p.m.