Please provide a reproducible .Rmd script to address the scenario below and produces all of your analysis and plots. Some reminders:
One hypothesis for why some plants are able to invade novel ecosystems is that they “escape” the effect of herbivory by moving to the new environment. Bull Thistle (Cirsium vulgare) is a exotic plant introduced to North America from Eurasia. In many parts of the United States it is considered a highly invasive noxious plant. In Nebraska however, it appears to be present but not reaching high abundances, even after many decades of establishment. There is a native thistle, the Tall Thistle (Cirsium altissimum), which flowers simultaneously with Bull Thistle. Many of the native herbivores found on Tall Thistles also attack Bull Thistle, and can consume up to 90% of all viable seeds. Could these native herbivores be suppressing invasive Bull Thistle populations, and could that suppression be related to distance from native Tall Thistle patches?
Hi! I’m a graduate student, and I set up an experiment testing the effects of distance to tall thistle patch on bull thistle seedling establishment density. I didn’t talk to a statistician before designing the experiment, which I know is a big no-no… So I know this is a big ask, but I really need your expert statistical advice! Here’s the info on my experiment:
I established \(4m^2\) plots of bull thistle seedlings at five distances (0, 7.5, 15, 30 and 60 meters) from a tall thistle patch. I did not attempt to control the presence or abundance of other patches of tall thistle at the site. At each distance, I established an insecticide (insect herbivory reduced) and control (insect herbivory present) plot. I sowed a fixed weight of Bull thistle seeds into the plots in April (538 seeds/ \(4m^2\)) and censused seedling density per \(1m2\) subplots the following September across all distances for insecticide and control plots. I repeated this experimental design in 10 sites in Washington and southern Benton counties. Additionally, I clumped sites into regions (4 regions), adding a second layer of spatial complexity. All sites within a region are closer to each other than in other regions. Region 1 is comprised of 3 sites at Healing Springs Natural Area, region 2 is a single site at Woolsey Wet Prairie, region 3 includes 3 sites Wilson Springs Preserve, and region 4 includes 3 sites at the Milo J. Shult Agricultural Research Center (the UA “farm”).
I tried a two way ANOVA for an initial analysis. But according to the ANOVA, there didn’t appear to be an effect of distance! This is devastating to the enemy release hypothesis, and I’m afraid I’m going to have to completely rewrite my thesis! I heard you just learned about mixed modeling. Could you please run a generalized or linear mixed model on my data and see if it picks up a different signal?!?
# List of packages necessary to run this script:
require(librarian, quietly = TRUE)
shelf(tidyverse,
MuMIn,
lme4,
DHARMa,
broom.mixed,
performance,
quiet = TRUE,
lib = tempdir())
# Set the web address where R will look for files from this repository
# Do not change this address
repo_url <- "https://raw.githubusercontent.com/LivingLandscapes/Course_EcologicalModeling/master/data/"
# Load data
thistles <- read.csv(paste0(repo_url,
"C_vulgare_seedling_dataset.csv"))
The top row is the insecticide (“i”) plots, bottom row is control plots (“c”). There are four points at each distance for the four subplots. You also need to setup the factors for region, site, plot and subplot thus:
Using these data, apply your knowledge of (generalized?) linear mixed models to determine if there is support for the enemy release hypothesis. Remember: not only are you analyzing the data, but you are also doing a post-mortem on the experiment/data collection/study design. Make sure to do model diagnostics and include any glaring issues you identify in the Results. See the Format instructions in the Overview section for required format and elements.