In dfalster/smatr3: (Standardised) Major Axis Estimation and Testing Routines

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  warning = FALSE
)

Below contains a brief overview on how to get users started with smatr

More detailed information about the package can be found here: Warton, D. I., Duursma, R. A., Falster, D. S., & Taskinen, S. (2012). smatr 3–an R package for estimation and inference about allometric lines. Methods in Ecology and Evolution, 3(2), 257-259.

First, install smatr`

# Install via CRAN or github
install.packages("smatr")
devtools::install_github("dfalster/smatr3")

then load the library

# Load library
library(smatr)

Note we will illustrate some of the core functions of smatr using one of the two data sets included in the package

data(leaflife)
data(leafmeas)

#For more details on each data set
?leaflife
?leafmeas

One sample analyses

Users can fit a single major axis to their data using the ma(y ~ x) or sma(y ~x) functions for y against x. The log argument allows quick transformations to one or both axes before estimation occurs. For example:

# Extract only low-nutrient, low-rainfall data from leaflife dataset:
leaf.low <- subset(leaflife, soilp == 'low' & rain == 'low')

# Fit a single MA for log(leaf longevity) vs log(leaf mass per area):
ma(longev ~ lma, log='xy', data=leaflife)

Testing for significance

Users can specify using the argument slope.test in the ma() or sma() function to test whether the estimated slope is significantly different from 1. Calling summary to sma or ma objects will return confidence intervals for the slope and/or elevation.

# Test if the MA slope is not significantly different from 1 for longevity and leaf mass per area (LMA):
ma_obj <- ma(longev ~ lma, log='xy', slope.test=1, data=leaflife) 

summary(ma_obj)
plot(ma_obj)

Multiple sample analyses

Often researchers wish to compare between slopes or intercepts of axes between different groups. Lets say we want to compare between high and low rainfall for low nutrient sites ONLY

# First we subset the data that is relevant
leaf.low.soilp <- subset(leaflife, soilp == 'low')

# Then using * we can fit SMAs separately at each of high and low rainfall sites and test for whether sites with different rainfall share a common slope
rain_sma_obj <- sma(longev~lma*rain, log="xy", data=leaf.low.soilp)

# Lets plot longevity vs LMA separately for each group first:
plot(rain_sma_obj)

By default, the sma() function will first fit a common slope for all groups and then test whether there is a difference in elevation

# Fit SMAs for each group and tests if several SMA lines share a common slope. 
rain_sma_obj <- sma(longev~lma*rain, log="xy", data=leaf.low.soilp)

summary(rain_sma_obj)

Again, using the slope.test will test whether the common slope is significantly different to 1.

# Fit SMAs separately at each of high and low rainfall sites,
# and test if there is a common slope and whether it is equal to 1:
rain_slope_obj <- sma(longev~lma*rain, log="xy", slope.test=1, data=leaf.low.soilp)

summary(rain_slope_obj)

The type argument allows users to test for a change in elevation

# Fit SMAs separately at each of high and low rainfall sites, and test whether sites differ in the elevation of their SMA
rain_elev_obj <- sma(longev~lma*rain, log="xy", type = "elevation", data=leaf.low.soilp)

summary(rain_elev_obj)

The type argument allows users to also test for shifts along a common axis

# Fit SMAs with common slope across each of high and low rainfall sites, and test for no shift along common SMA:
rain_shift_obj <- sma(longev~lma+rain, log="xy", type="shift", data=leaf.low.soilp)
summary(rain_shift_obj)