https://github.com/jebyrnes/divchangehmm

Development of hierarchical methods for examining change in species diversity
https://github.com/jebyrnes/divchangehmm

Last synced: 25 days ago
JSON representation

Development of hierarchical methods for examining change in species diversity

• Host: GitHub
• URL: https://github.com/jebyrnes/divchangehmm
• Owner: jebyrnes
• Created: 2015-06-08T22:00:16.000Z (over 9 years ago)
• Default Branch: master
• Last Pushed: 2016-03-08T22:08:18.000Z (over 8 years ago)
• Last Synced: 2024-06-11T17:33:53.140Z (5 months ago)
• Language: R
• Size: 24.7 MB
• Stars: 2
• Watchers: 8
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: Readme.Rmd

Awesome Lists containing this project

README

        
---

title: "Readme"

author: "Jarrett Byrnes"

date: "June 8, 2015"

output: md_document

---

```{r}

#Load methods for dealing with data

source("./R/dataGenerationFunctions.R")

#Load the fish data

fish <- read.csv("./data/rawData_SBCfish.csv")

fish <- subset(fish, fish$Year!=2000) #too much variation in this year

```

So, there are differences in # of years sampled from site to site and we want the core set that was sampled in all years filter to sites that are in the whole dataset

```{r}

fishSamples <- fish %>% group_by(Latitude, Longitude, Year) %>%

  summarise(n=n()) %>% ungroup() %>% 

  group_by(Latitude, Longitude) %>%

  summarise(nYearsSampled=n()) %>% 

  ungroup() %>%

  filter(nYearsSampled==max(nYearsSampled))

fishSamples$SampleID <- 1:nrow(fishSamples)

  

#use inner_join to filter down to those sites only

fish <- inner_join(fish, fishSamples)

#we'll need this later

nplots <- length(unique(paste(fish$Latitude, fish$Longitude)))

```

Create a simulated data set where for every scale, there are the same number of observations (total number of transects) and plot it relative to different predictors on the X-axis

Note, getSubData takes the following arguments:

  dataset - the name of a dataset 

  nplots - total number of unique plots

  n - the number of plots that will be used to create a n plot set

  noSpecies="-99999"- the code in the dataset for no species being in a plot

  uniquePerms=T - Only create unique permutations where no plots are shared within set of plots (sampling without replacement)

```{r, echo=FALSE}

simData <- lapply(2:17, function(m) getSubData(fish, nplots, m, sampleframe=fishSamples))

simData <- plyr::ldply(simData)

```

Look at it!

```{r plotTseries}

library(ggplot2)

qplot(Year, Aggregated_Richness, data=simData, group=SampleID, 

      geom="line", color=Bounded_region, facets=~Scale)

qplot(Scale, Aggregated_Richness, data=simData, group=Year, 

      geom="point",size=I(0), color=Year) +

  stat_smooth(fill=NA)

qplot(Bounded_region, Aggregated_Richness, data=simData, group=SampleID, 

      geom="point", color=Scale, facets=~Year)

```

And a sample analysis - although this will get better, right Forest?

```{r analysis}

library(lme4)

library(lmerTest)

#We want to do this, but it won't converge

#mod.full <- lmer(Aggregated_Richness ~ Year*Scale*Bounded_region +

#                   (1+ Year*Scale*Bounded_region|sampleID),

#                 data=simData)

#instead we do this

mod.novarslope.ranef<- lmer(Aggregated_Richness ~ Year*log(Scale)*Bounded_region +

                                   (1+log(Scale)|SampleID),

                                 data=simData)

summary(mod.novarslope.ranef)

```