library(metafor)
library(tidyverse)
Meta-Analysis Exercise: Does Grazing Exclusion Increase Plant Richness?
Introduction
You will extract data from 7 simulated ecological studies that all examine the effect of grazing exclusion on plant species richness.
Your task is to compute Hedges’ g for each study, then combine results in a meta-analysis using the metafor
package.
Load packages
Study 1: Temperate Grassland (USA)
In Minnesota, plant species richness averaged 14.2 (SD = 2.1) in grazed plots (n = 10) and 17.8 (SD = 2.4) in exclosures (n = 10).
Study 2: Mediterranean Pasture
In Spain, grazed areas (n = 15) had a mean richness of 12.1 (SD = 2.5), while exclosure plots had 14.7 species (SD = 2.2; n = 15).
Study 3: Alpine Meadow (Switzerland)
Richness in grazed plots was 13.8 (95% CI: 12.6–15.0; n = 12), compared to 16.3 (SD = 2.1; n = 12) in exclosures.
Hint: To convert CI to SD use:
# CI to SD: SD ≈ (upper - lower) / (2 * t)
<- #
ci_upper <- #
ci_lower <- qt(0.975, df = 11)# 1 less than n
t_crit
<- (ci_upper - ci_lower) / (2 * t_crit) sd
Study 4: Semi-arid Shrubland
Exclosure plots (n = 8) had 11.2 species on average (SD = 1.9). Grazed plots (n = 8) had 9.5 (SD = 2.0).
Study 5: African Savanna
Grazed plots (mean = 10.5, SE = 0.6, n = 10); Exclosures (mean = 13.2, SE = 0.5, n = 10)
Hint: Convert SE to SD: SD = SE * sqrt(n)
Study 6: Boreal Forest Edge
Exclosures: mean = 16.4, SD = 2.8, n = 9. Grazed: mean = 14.3, SD = 2.1. Sample sise for grazed plots not given, but design was fully balanced.
Study 7: Tropical Highlands
Species richness in grazed plots was 20.1 (SD = 2.5), while in exclosures it rose to 22.9 (SD = 3.1), with 11 plots in each treatment.
Build a Data Table
Use ?escalc
to see what the correct arguments are.
# Add your escalc results as rows
<- bind_rows(
all_es # escalc(measure = "SMD", ...)
)
Meta-Analysis & Forest Plot
<- rma(yi, vi, data = all_es) # change this to a suitable model type
res
forest(res, xlab = "Hedges' g (Effect of Grazing Exclusion on Richness)")