phenooooooooooooooooo.Rmd

---
title: "Untitled"
author: "Yara Ghabra 1006336056"
date: "2022-11-22"
output: html_document
---

```{r}
setwd("C:/Users/yarag_ldivedm/OneDrive/Documents/EEB313")
```




```{r}
library(tidyverse)
library(lmerTest)
library(lme4)
library(ggalt)
library(MuMIn)
library(sjmisc)
```


```{r}
#read in the unaltered phenophase data set
pheno_data<-read.csv("CCIN13215_20220120_tundra_phenology_database.csv")
#filter the dataset by flowering phenophase, control treatment, forb functional group, and species of interest (species that occur over the same time span with a lot of observations)
pheno_data_filt<-pheno_data %>% 
  filter(phenophase == "flower") %>% 
  filter(treatment == "CTL") %>% 
  filter(functional_group == "forb") %>% 
  filter(spp == "CARBEL" | spp == "DRALAC" | spp == "PAPRAD" | spp == "POLVIV" | spp == "PEDHIR" | spp == "SAXOPP" | spp == "SILACA" | spp == "STECRA")
#seeing how many species occur in both wet and dry
numbers<-pheno_data_filt %>% 
  group_by(spp,soil_moisture) %>% 
  summarise(n=n()) %>% 
  ungroup() %>% 
  filter(soil_moisture == "wet" | soil_moisture =="dry" | soil_moisture == "moist")
  
#filtering by three species of interest - occur in both wet and dry, have a similar time range of observations, and have a lot of observations
spp_filt<-pheno_data_filt %>% 
  filter(spp == "DRALAC" | spp == "PAPRAD" | spp == "POLVIV" | spp == "PEDHIR") %>% 
  filter(soil_moisture =="wet" | soil_moisture =="moist" | soil_moisture =="dry")
spp_filt[spp_filt == "moist"] <- "wet"

#only including wet and dry 
spp_filt[spp_filt == "moist"] <- "wet"
```


```{r}
#creating a dataset per species for t tests
polviv_ttest<-spp_filt %>% 
  filter(spp == "POLVIV")
dralac_ttest<-spp_filt %>% 
  filter(spp == "DRALAC")
paprad_ttest<-spp_filt %>% 
  filter(spp == "PAPRAD")
pedhir_ttest<-spp_filt %>% 
  filter(spp == "PEDHIR")
polviv_ttest[polviv_ttest == "moist"] <- "wet"
dralac_ttest[dralac_ttest == "moist"] <- "wet"
paprad_ttest[paprad_ttest == "moist"] <- "wet"
pedhir_ttest[pedhir_ttest == "moist"] <- "wet"
#removing blank values from the POLVIV dataset
polviv_ttest<-polviv_ttest %>% 
  filter(soil_moisture == "dry" | soil_moisture == "wet")
```


###check assumptions for t tests!!!!!!
two assumptions:
1. The dependent variable must be normally distributed in both samples
2. The variance of the dependent variable must be approximately equal between the two samples

Shapiro-wilk's test:

in this case if we get a non-significant p-value, our data is normally distributes

```{r}
#Shapiro-wilk's test to test normality
shapiro.test(paprad_ttest$DOY)
#not normally distributed
shapiro.test(polviv_ttest$DOY)
#not normally distributed
shapiro.test(pedhir_ttest$DOY)
#not normally distributed
shapiro.test(dralac_ttest$DOY)
#normally distributed

# F-test to test for homogeneity in variances
# H0: Ratio of variance is equal to 1, i.e., samples have equal variance
var.test(DOY ~ soil_moisture, data=paprad_ttest)
var.test(DOY ~ soil_moisture, data=polviv_ttest)
var.test(DOY ~ soil_moisture, data=pedhir_ttest)
var.test(DOY ~ soil_moisture, data=dralac_ttest)
```


```{r}
#performing t test for paprad
t.test(DOY~soil_moisture, data = paprad_ttest)
```

```{r}
#t test for polviv
t.test(DOY~soil_moisture, data = polviv_ttest)
```

```{r}
# t test for dralac
t.test(DOY~soil_moisture, data = dralac_ttest)
```


```{r}
#t test for pedhir
t.test(DOY~soil_moisture, data = pedhir_ttest)
```

```{r}
#attempting a t test with one site per species (site with the most observations)
#polviv shows a significant difference at both sites which contain both dry and wet soils, however the results are opposite
polviv_ttest2<-polviv_ttest %>% 
  filter(study_area == "Endalen")
t.test(DOY~soil_moisture, data = polviv_ttest2)

polviv_ttest3<-polviv_ttest %>% 
  filter(study_area == "Adventdalen")
t.test(DOY~soil_moisture, data = polviv_ttest3)
```


```{r}
#creating a linear model to test the effects of soil moisture on flowering date
#random intercept, fixed slope
mix_model_int<-lmer(DOY~soil_moisture * spp +
                      (1|study_area/subsite) + (1|year),
                    data = spp_filt)
summary(mix_model_int)
```

###check assumptions for linear model!!!!!
1. Normality at each X value
2. Homogeneity of variances at each X
3. Independence of observations

```{r}
library(effects)
library(sjPlot)
library(glmmTMB)
plot_model(mix_model_int, type='diag')
```

```{r}
#looking at the effects without species
mix_model_int2<-lmer(DOY~soil_moisture +
                      (1|study_area/subsite) + (1|year),
                    data = spp_filt)
summary(mix_model_int2)
```

```{r}
#model selection
AICc(mix_model_int, mix_model_int2)
```

```{r}
#simple visualizations of our data

spp_filt %>% 
  ggplot(aes(x=soil_moisture, y=DOY, colour=spp))+
  geom_boxplot()+
  labs(x="Soil Moisture", y="Flowering Day")

```