CABIN Benthic_data.Rmd

# CABIN Data Exploration

Preliminary wrangling, tidying and exploration for benthic data within all regions sampled in Canada within the Canadian Aquatic Biomonitoring Network (CABIN). 

CABIN is an aquatic biomonitoring program for freshwater ecosystems that began in 1987. Sites are divided by "Reference", or "Test" sites.

Reference sites represent sites without industrial, agriculture, or development activity. Test sites may or may not have similar activity, and data from the reference sites is used to assess changes to macroinvertebrate communities, water chemistry, and habitat. 

This data set is a collection of 30 or so spreadsheets from different regions, with different site visit identifiers and variables across each type of spreadsheet ("study", "habitat" and "benthic"). This is an example of how to aggregate and sort somewhat disparate data from different sources for graphing and exploration.

The differences between test and reference sites are graphed here for the initial analysis to begin to see if a) we can detect a difference in pH, macroinvertebrate abundance, and percent sorted between test and reference sites. Further analyses are coming, and will be done with the MVAbund package in R for multivariate environmental data. 

Data retrieved from the open source data set: https://open.canada.ca/data/en/dataset/13564ca4-e330-40a5-9521-bfb1be767147

Open-sourced R packages used: tidyverse, tidyhydat, zoo, lubridate, ggplot, scales, data.table

source code with quality control methods for this dataset:  https://github.com/Jacqui-123/CABIN/blob/main/CABIN%20Benthic_data.Rmd

```{r, echo = FALSE, include = FALSE}
library(png)
library(dplyr)
library(tidyverse)
library(reticulate)
library(zoo)
library(lubridate)
library(ggplot2)
library(data.table)
library(hexbin)
library(knitr)
library(sqldf)
library(scales)

```

### 1) Load in the data
Smoosh all the benthic datasheets into one dataframe, with a column of region to be able to keep track of the different regions each data set originally came from. Do the same thing for the "study" and "habitat" spreadsheets.
```{r}
library(data.table)
files <- list.files(path = "/Users/LevyJ/Desktop/CABIN", pattern = "^cabin_benthic.*?\\.csv", full.names = TRUE) 
df_benthic <- map_df(files, ~read_csv(., col_types = cols(.default = "c")), .id = "Region", encoding = 'Latin-1') 
#do this for all files
```

```{r, echo = FALSE, include = FALSE}
library(data.table)
files <- list.files(path = "/Users/LevyJ/Desktop/CABIN", pattern = "^cabin_study.*?\\.csv", full.names = TRUE) 
df_study <- map_df(files, ~read_csv(., col_types = cols(.default = "c")), .id = "Region", encoding = 'Latin-1') 
```

```{r, echo = FALSE, include = FALSE}
library(data.table)
files <- list.files(path = "/Users/LevyJ/Desktop/CABIN", pattern = "^cabin_habitat.*?\\.csv", full.names = TRUE) 
df_habitat <- map_df(files, ~read_csv(., col_types = cols(.default = "c")), .id = "Region", encoding = 'Latin-1') 
```

### 2) Data Tidying 
a\) Benthic Data

-Rename the columns to English and standard characters to be easier to work with. (code not shown)
```{r, echo = FALSE, include = FALSE}
df_benthic <- df_benthic %>%
  rename("Site_Visit_ID" = 'SiteVisitID/IdentifiantdeVisite',
         "Site" = "Site/Site",
         "Year" = "Year/Année",
         "Sampling_Device" = "SamplingDevice/Dispositifd'échantillonnage",
         "Kick_Time" = "KickTime/Périodedelapassedufilettroubleau",
         "Mesh_size" = "MeshSize/Maillage",
         "Subsample" =	"SubSample/Sous-échantillon",
         "Julian_day" ="JulianDay/JourJulien",
         "Total_sample" ="TotalSample/Échantillontotal",
         "Status" = "Status/État",
         "Taxonomist" = "Taxonomist/Taxonomiste",
         "organization" = "Organization/Organisation",
         "address" = "Address/Adresser",
         "city" = "City/Ville",
         "province" = "Province/Province",
         "phylum" = "Phylum/Phylum",
         "class" = "Class/Classe",
         "Order" = "Order/Ordre",
         "Family" = "Family/Famille",
         "Genus" = "Genus/Genre",
         "Species" = "Species/Espèce",
         "Replicate" = "Replicate/Réplicat",
         "Count" = "Count/Décompte",
         "ITIS" ="ITIS_TSN",
         "Valid" = "Valid/Valide")
```
-Create a new column “unique ID” with the site visit ID, site, and year separated by a “-”. This is so we have a unique id for each site visit and year. This was done for each set of spreadsheet to create a unique identifier for each site visit that can be used across all types of datasheets.
```{r}
df_benthic <- df_benthic %>%
  tidyr::unite("Unique_Id", c("Site_Visit_ID", "Site", "Year" ), sep = "_", remove = TRUE) 
```

-Select only kick net samples, sample # = 1, mesh size 400, and calculate the total abundance for each unique site. Individual taxa are removed for now. 
```{r, echo = FALSE, include = FALSE}
#change count column to numeric data
df_benthic$Count = as.numeric(as.character(df_benthic$Count) )
df_benthic$Count <- round (df_benthic$Count, digits = 0)

df_benthic <- df_benthic %>%
 #filter(Sampling_Device == 'Kick Net') %>%
  #filter(Mesh_size == "400") %>%
   filter(Sample_Number == '1') %>%
  group_by(Unique_Id) %>%
  mutate("Abundance" = sum(Count) ) %>%
  #removing the individual taxa for now, just using total abundance
  select(Region, Unique_Id, Julian_day, Sampling_Device, Sample_Number, Kick_Time, Mesh_size, Sample_Number, Subsample, Total_sample, Status, Abundance) %>%
 distinct()

```

b\) Study data tidying:
-Create a unique site identifier to be able to accurately match up data
-Keep only the "River" sites, as some are lake or wetland sites.

```{r, echo = FALSE, include = FALSE}
#-(Study data columns had to be renamed in excel)
#-Mutate a new column "unique ID" with the site visit ID, site, and year separated by a "-" (code not shown, see above)

df_study <- df_study %>%
   #some sites have type = lake or wetland, we want to remove these
  #filter(Type == 'River') %>%
  tidyr::unite("Unique_Id", c("SiteVisitID", "Site", "Year"), sep =  "_", remove = TRUE) 

```

c\) Habitat data tidying: 
 
-Create a unique site identifier to be able to accurately match up data

```{r, echo=FALSE, include=FALSE}
#-Rename to make the habitat data easier to work with.
#Unit, year, sample_number were renamed in excel.

df_habitat <- df_habitat %>%
  rename("Site_Visit_Id" = 'SiteVisitID/IdentifiantdeVisite',
         "Site" = "Site/Site",
         "Julian_day" = "JulianDay/JourJulian",
         "Status" ="Status/État",
         "Stat_Num" = "status",
         "Type" = "Type/Type",
         "Value" = "Value/Valeur",
         "Computed" = "Computed/Calculé",
         "Notes" = "Note/Remarque" ) %>%
  #deal with status being 1 in this dataset but reference or test etc in the others
  mutate(Status = replace(Status, Stat_Num == "1", "Reference" ),
         Status = replace(Status, Stat_Num =="2", "Test"),
         Status = replace(Status, Stat_Num == "3", "Potential Reference")) %>%
  dplyr::select(Region, Site_Visit_Id, Site, Protocol, Julian_day, Year, Sample_Number, Status, Stat_Num, QAQC, Type, Variable, VariableDescription, Unit, Value, MDL, Computed, Notes)
```


```{r, echo= FALSE, include=FALSE}
#-removed the 65 NAs for Protocol and Site - these are where the previous line from MDL/notes/computed #continued on the next line and entered "<0.0005 mg/L" etc for site. Deleted these, but might be missing #some data for these sites - 

#*NOTE: can delete valus from hab #5 spreadsheet with all the weird values for site - I confirmed that it doesn't delete data.**
df_habitat <-df_habitat %>%
  #drop_na(Protocol)  %>%
tidyr::unite("Unique_Id", c("Site_Visit_Id", "Site", "Year" ), sep = "_", remove = TRUE )

```

-Pivot habitat data so the variables are in columns and the values in rows 

```{r}

df_habitat <- df_habitat %>%
#keep only first sample from triplicates
  filter(Sample_Number == "1") %>%
  #get rid of protocols other than wadeable streams
  #filter(Protocol == "CABIN - Wadeable Streams") %>% 
  #keep only alkalinity and pH data
  filter(Variable == "General-Alkalinity" | Variable == "General-pH")  %>%
  filter(Type == "Water Chemistry")%>%
  #Value needs to be as.numeric or pivot won't work. 
  mutate_at(c('Value'), as.numeric) %>%
  #unite so have units of measurement in column name as well when pivot
  tidyr::unite("Habitat Variable", c("Variable", "Unit"), sep = "_") %>%
  select(-c( "MDL", "VariableDescription")) %>%
  #pivot, preserving multiple columns (value, notes, computed)
  pivot_wider(names_from = "Habitat Variable", values_from = c("Value", "Notes", "Computed") ) %>%
  rename("pH" = "Value_General-pH_pH",
         "Alkalinity" = "Value_General-Alkalinity_mg/L") %>%
    distinct()

```

### 3) Merge data together

-Change df to Data.tables for quicker processing speed using the data.table vignette
```{r, echo = FALSE, include = FALSE}
setDT(df_benthic)
setDT(df_study)
setDT(df_habitat)
```

-Join the three data tables together. 

```{r}

#merge on benthic and habitat data, keeping everything: 

df_both<- merge(df_habitat, df_benthic, by = "Unique_Id", all = TRUE)

#merge df_both and the study data - left join to keep only the df_both (hab + benth) sites, as study data doesn't have any new variables of interest.  

df_all <- merge(df_both, df_study, by = "Unique_Id", all.x = TRUE)

#length(unique(df_all$Unique_Id)) #sanity check, should be 12408

```

Tidy the joined df to get rid of lake/wetland/non wadeable stream sites 

```{r}
#tidy to get rid of wetlands, lakes, etc- make sure to keep NAs
df_all <- df_all %>%
  filter(is.na(Protocol) | Protocol == "CABIN - Wadeable Streams") %>%
  filter(is.na(Sampling_Device) | Sampling_Device == "Kick Net") %>%
  #note: we are keeping if River or wadeable
  filter(is.na(Type.y) | Type.y  == "River") %>%
  filter(is.na(Mesh_size) | Mesh_size == "400") %>%
  mutate(Status = coalesce(Status.x, Status.y)) %>%
  mutate(Region = coalesce(Region.x, Region.y))
```

```{r, include = FALSE}
#Note: many sites don't have Protocol listed at all. For those, I kept them if "Type" == River or NA, and then Went through this file with Wendy Monk. Borrow pit (1 site) removed. For sites with notes and site name, these are all rivers/streams and can be kept. Sites without any identifying parameters to be deleted (code below).

#Protocol_Nas <- df_all[is.na(df_all$Protocol),]
#write.csv(Protocol_Nas, "No_Protocol_Sites.csv")

```

```{r, echo = FALSE}

#for sites without a Protocol, get rid of the sites that also don't have Type == River.
df_all <- df_all %>%
#remove borrow pit, it's a pit...not a river...
filter(LocalBasinName != "Borrow Pit" | is.na(LocalBasinName)) %>%
filter(!is.na(Protocol) | !is.na(Type.y)) #removes 70 sites
```

```{r, echo = FALSE, include=FALSE}
#Sanity check after merging - count number of test, potential test, and reference sites there are. These should add to 6987 (they do!) and there should be no NAs for Status column

length(unique(df_all$Unique_Id)) #6987

num_ref_sites <- sqldf( "select Status, Unique_Id
                     from df_all Where Status Like '%Refere%' ") #3448


num_test_sites <- sqldf("select Status, Unique_Id
                         from df_all where Status LIKE 'Test' ") #3539

#3448 + 3539 = 6987

#sum(is.na(df_all$Status)) #Check that there are No Nas for "Status"
rm(num_ref_sites, num_test_sites)
```



```{r, echo = FALSE, include = FALSE}
#-Load in the region names and add to the merged dfs using the region names file (file made in excel)
Region_names <- read.csv("Region_names.csv", colClasses =c("character", "character", "character")) 
df_all <- full_join(df_all, Region_names, by = 'Region') 

df_all <- df_all %>%
  select(-c(File_Name))

rm(Region_names)

#sum(is.na(df_all$Region_Name)) #should be no NAs for Region_Name
    
```

### 4) Get ready to graph 

-Tidy and reformat df_all pH and alkalinity data so that it is as numeric, and has the correct number of digits
```{r, echo = FALSE, include = FALSE}

setDT(df_all)

df_all$`pH` = as.numeric(as.character(df_all$`pH`) )
df_all$`pH` <- round(df_all$`pH`, digit = 3)

df_all$`Alkalinity` = as.numeric(as.character(df_all$`Alkalinity`) )
df_all$`Alkalinity` <- round(df_all$`Alkalinity`, digit = 4)

df_all <- df_all %>%
  mutate_at(c('Subsample'), as.numeric)

```

-Graphing prep: make separate data.table for test, reference, and all sites for quicker processing
```{r, echo= FALSE, include = FALSE}

sum(is.na(df_all$Status)) #double check no Nas for status

df_ref <- df_all %>%
  filter(Status == "Reference" | Status == "Potential Reference") %>%
  select(Unique_Id, Region_Name, Protocol, Type.y, Status, Abundance, Subsample, pH, Alkalinity ) %>%
  #filter(str_detect(Unique_Id_Sample, "_1$")) %>%
  mutate_at(c('Subsample'), as.numeric) %>%
  distinct()

df_test <- df_all %>%
  filter(Status == "Test" ) %>%
  select(Unique_Id, Region_Name, Protocol, Type.y, Status, Abundance, Subsample, pH, Alkalinity ) %>%
  #filter(str_detect(Unique_Id_Sample, "_1$")) %>%
  mutate_at(c('Subsample'), as.numeric) %>%
  distinct()

df_status <- df_all %>% 
  filter(Status == "Test" | Status == "Reference" | Status == "Potential Reference") %>%
  select(Unique_Id, Region_Name, Protocol,Type.y, Status, Abundance, Subsample, pH, Alkalinity ) %>%
  #filter(str_detect(Unique_Id_Sample, "_1$")) %>%
  mutate_at(c("pH"), as.numeric) %>%
  mutate_at(c('Subsample'), as.numeric) %>%
  distinct()

```

-Make a boxplot function so that it's easier to make multiple graphs with different inputs
```{r}
#Graphing function: boxplot 
boxplot <- function(data, x, y, fill, title, ylabtitle) {
  p <-
    ggplot(data, aes({{x}}, {{y}}, fill = {{fill}} )) +
geom_boxplot() +
  theme_bw() +
  geom_point() +
    theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
   stat_boxplot(geom = "errorbar", width = 0.25) +
  ggtitle({{title}}) +
   ylab({{ylabtitle}}) +
 theme(legend.position = 'none') +
scale_x_discrete(labels = wrap_format(10)) +
  viridis::scale_fill_viridis(discrete = TRUE)
    print(p)
}

```

### 5) Graphing: Percent Sorted by reference and test sites

```{r, echo=FALSE, fig.align='center', warning=FALSE}

#b) ref sites
boxplot(df_ref,Region_Name, Subsample, Region_Name, "Percent Sorted by Region \n(Potential Reference or Reference Sites)", "Percent Sorted" )

```

```{r, echo=FALSE, fig.align='center', warning= FALSE}

#c) Test sites
df_test <- df_test %>% filter(Subsample <= 100)
boxplot(df_test,Region_Name, Subsample, Region_Name, "Percent Sorted by Region \n(Test Sites)", "Percent Sorted" )


#lost Southwestern Hudson Bay Drainage area - makes sense, there are only 3 test sites & they don't have Abundance data

```
```{r, include = FALSE, echo = FALSE}
#calculate how many sites for test and reference 

x = sum(is.na(df_ref$Abundance))
y = length(unique(df_ref$Unique_Id) )

m = sum(is.na(df_test$Abundance))
n = length(unique(df_test$Unique_Id) )

as.numeric(c(x,y, m,n))

y-x

n-m
```

Table with # of data points for each region (ref/potential ref sites):

```{r, echo=FALSE, warning= FALSE}

df_ref %>%
  filter(Subsample >= 1 ) %>%
  count(Region_Name)

```
Table with # of data points for each region (test sites) 

```{r, echo=FALSE, warning= FALSE}
 df_test %>%
  filter(Subsample >= 1 ) %>%
  count(Region_Name)

```

### 6) Graphing: Total Abundance (macroinvertebrate count) data

**Note: looks like there may be outliers? These are not excluded from the scatterplots below**
```{r, echo = FALSE, fig.align='center', warning=FALSE}

#a) boxplots, reference sites
boxplot(df_ref,Region_Name, Abundance, Region_Name, "Abundance by Region \n(Reference Sites)", 'Total Abundance (count)' )

```

```{r, echo = FALSE, fig.align='center', warning=FALSE}
#b) boxplots, test sites

boxplot(df_test,Region_Name, Abundance, Region_Name, "Abundance by Region \n(Test Sites)", 'Total Abundance (count)' )

```
Abundance data, # of data points for each region (ref sites) 

```{r, echo=FALSE, warning= FALSE}

df_ref %>%
  filter(Abundance >= 1 ) %>%
  count(Region_Name)
```
Abundance data, # of data points for each region (test sites) 

```{r, echo=FALSE, warning= FALSE}
 df_test %>%
  filter(Abundance >= 1 ) %>%
  count(Region_Name)

```


```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}

#c) Scatterplots: Abundance vs percent sorted, by region - Reference sites
#(y=300 dotted line)
df_ref %>%
  filter(Subsample <= 100) %>%
ggplot(mapping = aes(x = Subsample, y = Abundance, colour = Region_Name) ) +
  geom_point(size=2) + 
  geom_hline(yintercept = 300, linetype = "dotted") +
facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 20))) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
  theme(strip.text = element_text(size=7)) +
  ggtitle( "Percent Sorted vs Abundance \n (Reference/Potential Reference Sites) ") +
 ylab("Total Abundance (count) ") +
 xlab("Percent Sorted") + 
  theme(legend.position= 'none') +
    viridis::scale_colour_viridis(discrete = TRUE)

```



```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}
#d) Scatterplots: Abundance vs percent sorted, by region - Test sites
#(y=300 dotted line)
df_test %>%
  filter(Subsample <= 100) %>%
ggplot(mapping = aes(x = Subsample, y = Abundance, colour = Region_Name) ) +
  geom_point(size=2) + 
  geom_hline(yintercept = 300, linetype = "dotted") +
facet_wrap(~Region_Name) +
  facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
    theme(strip.text = element_text(size=7)) +
  ggtitle( "Percent Sorted vs Abundance \n (Test Sites) ") +
 ylab("Total Abundance (count) ") +
 xlab("Percent Sorted") + 
  theme(legend.position= 'none') +
    viridis::scale_colour_viridis(discrete = TRUE)

```


```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}
#e) Scatterplot: Abundance vs percent sorted,for all regions together
#(dotted line at y = 300)
df_status %>%
  filter(Subsample <= 100) %>%
  mutate(Status = replace(Status, Status == "Potential Reference", "Reference" )) %>%
ggplot(mapping = aes(x = Subsample, y = Abundance, colour = Status) ) +
  geom_point(size=3, alpha = 0.3) + 
  geom_hline(yintercept = 300, linetype = "dotted") +
facet_wrap(~Status) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
  ggtitle( "Percent Sorted vs Abundance \n (Test & Reference/Potential Reference Sites) ") +
 ylab("Total Abundance (count) ") +
 xlab("Percent Sorted") +
  theme(legend.position= 'none') +
      viridis::scale_colour_viridis(discrete = TRUE)

```

### 7) Graphing: pH data (& Percent sorted)

```{r,    echo = FALSE, fig.align='center', warning=FALSE}
#a) Boxplots: Reference sites

boxplot(df_ref, Region_Name, pH, Region_Name, "pH by Region \n(Reference and Potential Reference Sites)", 'pH')

```

```{r,    echo = FALSE, fig.align='center', warning=FALSE}
#b) Boxplots: Test sites
#(outlier pH = 14 was not removed)
df_test <- df_test %>% filter(pH < 14)
boxplot(df_test, Region_Name, pH, Region_Name, "pH by Region \n(Test Sites)", 'pH')
```

pH data, # of data points for each region (ref sites) 
```{r, echo=FALSE, warning= FALSE}
 df_ref %>%
  filter(pH >= 1 ) %>%
  count(Region_Name)
```

pH data, # of data points for each region (test sites) 
```{r, echo=FALSE, warning= FALSE}
 df_test %>%
  filter(pH >= 1 ) %>%
  count(Region_Name)
```


```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}

#c) Scatterplots: pH vs percent sorted 
#Reference sites
#(Outliers have not been removed)
df_ref %>%
filter(Subsample <= 100) %>%
    na.omit() %>%
ggplot(mapping = aes(x = pH, y = Subsample, colour = Region_Name )) +
  geom_point(size=2) +
  facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
 theme_bw() +
    theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
      theme(strip.text = element_text(size=7)) +
  ggtitle( "pH vs Percent Sorted \n (Reference or Potential Reference Sites) ") +
 ylab("Percent Sorted") +
 xlab("pH") +
theme(legend.position = "None") +
        viridis::scale_colour_viridis(discrete = TRUE)
```


```{r, fig.width = 5, fig.height= 5,  echo = FALSE, fig.align='center'}
#d) Scatterplots: Percent sorted vs pH  
#Test sites
#(Outliers have not been removed)
df_test %>%
filter(Subsample <= 100) %>%
    filter(pH <= 14) %>%
    na.omit() %>%
ggplot(mapping = aes(x = pH, y = Subsample, colour = Region_Name )) +
  geom_point(size=2) +
  facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
 theme_bw() +
    theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
      theme(strip.text = element_text(size=7)) +
  ggtitle( "pH vs Percent Sorted \n (Test Sites) ") +
 ylab("Percent Sorted") +
 xlab("pH") +
theme(legend.position = "None") +
        viridis::scale_colour_viridis(discrete = TRUE)

```


```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}
#e) Scatterplot: Percent sorted vs pH  -- all regions
#(Outliers >= pH 14 removed but no others removed)

df_status %>%
  filter(Subsample <= 100) %>%
    filter(pH <= 14) %>%
  mutate(Status = replace(Status, Status == "Potential Reference", "Reference" )) %>%
ggplot(mapping = aes(x = pH, y = Subsample, colour = Status) ) +
  geom_point(size=3, alpha = 0.3) + 
facet_wrap(~Status) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
  ggtitle( "pH vs Percent Sorted \n (Test and Refer. Sites for all regions) ") +
 ylab("Percent Sorted ") +
 xlab("pH") +
  theme(legend.position= 'none') +
        viridis::scale_colour_viridis(discrete = TRUE)

```

### 8) Graphing: pH data (& Abundance)
```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}
#a) Reference sites 

df_ref %>%
filter(Subsample <= 100) %>%
    na.omit() %>%
ggplot(mapping = aes(x = pH, y = Abundance, colour = Region_Name )) +
  geom_point(size=2) +
  facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
 theme_bw() +
    theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
        theme(strip.text = element_text(size=7)) +
  ggtitle( "pH vs Abundance \n (Reference or Potential Reference Sites) ") +
 ylab("Total Abundance") +
 xlab("pH") +
theme(legend.position = "None") +
          viridis::scale_colour_viridis(discrete = TRUE)

```

```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}
#b)Test Sites

df_test %>%
filter(Subsample <= 100) %>%
    na.omit() %>%
      filter(pH <= 14) %>%
ggplot(mapping = aes(x = pH, y = Abundance, colour = Region_Name )) +
  geom_point(size=2) +
  facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
 theme_bw() +
    theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
        theme(strip.text = element_text(size=7)) +
  ggtitle( "pH vs Abundance \n (Test Sites) ") +
 ylab("Total Abundance") +
 xlab("pH") +
theme(legend.position = "None") +
          viridis::scale_colour_viridis(discrete = TRUE)

```

```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}
#C) pH vs Abundance, all regions

df_status %>%
  filter(Subsample <= 100) %>%
    filter(pH <= 14) %>%
  mutate(Status = replace(Status, Status == "Potential Reference", "Reference" )) %>%
ggplot(mapping = aes(x = pH, y = Abundance, colour = Status) ) +
  geom_point(size=3, alpha = 0.3) + 
facet_wrap(~Status) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
  ggtitle( "pH vs Abundance \n (Test and Refer. Sites for all regions) ") +
 ylab("Abundance ") +
 xlab("pH") +
  theme(legend.position= 'none') +
            viridis::scale_colour_viridis(discrete = TRUE)

```

### 9) Graphing: Alkalinity data & Percent Sorted

**Note: Lower (<20.0 mg/L) values differ depending on the lab and might not be entirely accurate to compare across the sites. For example in some cases they list a "low quantification limit" with a value close to zero (values = .035 for example). Or an MDL threshold like <20.0, or <10.0, and in each case is entered into CABIN as 1/2 the MDL (ie 10.0, or 5.0). All of these values were retained, as is, for the plots down below. 

```{r, echo = FALSE, warning=FALSE }

df_ref_alk <- df_ref %>%
  filter(Alkalinity >= 0)
boxplot(df_ref_alk, Region_Name, Alkalinity, Region_Name, "Alkalinity by Region \n(Reference and Potential Reference Sites)", 'Alkalinity (mg/L)')

#sum(is.na(df_ref$Alkalinity)) #1456 NAs

```

```{r, echo= FALSE, warning = FALSE}

boxplot(df_test, Region_Name, Alkalinity, Region_Name, "Alkalinity by Region \n(Test Sites)", 'Alkalinity (mg/L)')

#sum(is.na(df_test$Alkalinity))
```

Alkalinity data, # of data points for each region (ref sites) 

```{r, echo = FALSE}
 df_ref %>%
  filter(Alkalinity >= 1 ) %>%
  count(Region_Name)
```

Alkalinity data, # of data points for each region (test sites) 

```{r, echo = FALSE}
df_test %>%
  filter(Alkalinity >= 1) %>%
  count(Region_Name)

```


```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}

#c) Scatterplots: Alkalinity vs percent sorted, by region - Reference sites
#(y=300 dotted line)
df_ref %>%
  filter(Subsample <= 100) %>%
  filter(Alkalinity >=1) %>%
ggplot(mapping = aes(x = Alkalinity, y = Subsample, colour = Region_Name) ) +
  geom_point(size=2) + 
 # geom_hline(yintercept = 300, linetype = "dotted") +
facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
  theme(strip.text = element_text(size=7)) +
  ggtitle( "Percent Sorted vs Alkalinity \n (Reference/Potential Reference Sites) ") +
 ylab("Percent Sorted") +
 xlab("Alkalinity (mg/L)") + 
  theme(legend.position= 'none') +
    viridis::scale_colour_viridis(discrete = TRUE)

```


```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}

#d) Scatterplots: Alkalinity vs percent sorted, by region - Test sites
#(y=300 dotted line)
df_test %>%
  filter(Subsample <= 100) %>%
  filter(Alkalinity >=1) %>%
ggplot(mapping = aes(x = Alkalinity, y = Subsample , colour = Region_Name) ) +
  geom_point(size=2) + 
 # geom_hline(yintercept = 300, linetype = "dotted") +
facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
  theme(strip.text = element_text(size=7)) +
  ggtitle( "Percent Sorted vs Alkalinity \n (Test Sites) ") +
 ylab("Percent Sorted") +
 xlab("Alkalinity (mg/L)") + 
  theme(legend.position= 'none') +
    viridis::scale_colour_viridis(discrete = TRUE)

```

### 10) Graphing: Alkalinity data & Abundance

**Note: MDL values included
```{r, fig.width = 5, fig.height= 5, warning= FALSE,   echo = FALSE, fig.align='center'}

#a) Scatterplots: Alkalinity vs Abundance, by region - Reference sites
#(y=300 dotted line)
df_ref %>%
  filter(Alkalinity >=1) %>%
ggplot(mapping = aes(x = Alkalinity, y = Abundance, colour = Region_Name) ) +
  geom_point(size=2) + 
 geom_hline(yintercept = 300, linetype = "dotted") +
facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
  theme(strip.text = element_text(size=7)) +
  ggtitle( "Total Abundance vs Alkalinity \n (Reference/Potential Reference Sites) ") +
 ylab("Total Abundance (count)") +
 xlab("Alkalinity (mg/L) ") + 
  theme(legend.position= 'none') +
    viridis::scale_colour_viridis(discrete = TRUE)

```

```{r, fig.width = 5, fig.height= 5,    echo = FALSE, fig.align='center'}
#b) Scatterplots: Alkalinity vs Abundance, by region - test sites

df_test %>%
  filter(Alkalinity >=1) %>%
ggplot(mapping = aes(x = Alkalinity, y = Abundance, colour = Region_Name) ) +
  geom_point(size=2) + 
  geom_hline(yintercept = 300, linetype = "dotted") +
facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
  theme(strip.text = element_text(size=7)) +
  ggtitle( "Total Abundance vs Alkalinity \n (Test Sites) ") +
 ylab("Total Abundance (count)") +
 xlab("Alkalinity (mg/L) ") + 
  theme(legend.position= 'none') +
    viridis::scale_colour_viridis(discrete = TRUE)


```


```{r, fig.width = 5, fig.height= 5, warning = FALSE, echo = FALSE, fig.align='center'}
#c) Alkalinity vs Abundance, all regions

df_status %>%
  mutate(Status = replace(Status, Status == "Potential Reference", "Reference" )) %>%
ggplot(mapping = aes(x = Alkalinity, y = Abundance, colour = Status) ) +
  geom_point(size=3, alpha = 0.3) + 
facet_wrap(~Status) +
      theme_bw() +
 theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
  ggtitle( "Alkalinity vs Abundance \n (Test and Refer. Sites for all regions) ") +
 ylab("Total Abundance ") +
 xlab("Alkalinity (mg/L)") +
  theme(legend.position= 'none') +
            viridis::scale_colour_viridis(discrete = TRUE)


```


```{r, echo = FALSE, include = FALSE}


#MDL unique notes examples: 
#Data was converted from '<20.0' to '10', Data below the MDL was entered into CABIN as 1/2 the MDL --> means they entered 10.

#Data was converted from '<10' to '5', Data below the MDL was entered into CABIN as 1/2 the MDL --> means they entered 5

#Issue: there are multiple MDL's 
#sometimes MDL is listed, sometimes it's not- 

#sometimes data just has a "low quantification limit", and we see values close to 0.

#solution: just flag them. Most of them are below 10 which is all basically the same in terms of buffering capacity
  
```



```{r, echo = FALSE, include = FALSE}

#ignore, this is old code 
#cbPalette <- c("#999999","#CC79A7" , "#E69F00", "#56B4E9", "#009E73", "#F0E442", "#0072B2", "#332288", "#D55E00", "#44AA99")
  #  scale_fill_manual(values=cbPalette)
  #scale_colour_manual(values = c("#999999","#CC79A7" , "#E69F00", "#56B4E9", "#009E73", "#F0E442", "#0072B2", "#332288", "#D55E00", "#44AA99")) +



### Trying to get scatterplots into a function.... 

#scatplot <- function(data, x, y, facetby, title, xlabtitle, ylabtitle) {
 # p <- ggplot(data, aes({{x}}, {{y}})) +
#  geom_point(size=2) +
 #     geom_hline(yintercept = 300, linetype = "dotted") +
  #  facet_wrap(enquo(facetby)) +
   #     theme_bw()  +
    # theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
#ggtitle( {{title}}) +
# xlab({{xlabtitle}}) + 
 #ylab({{ylabtitle}}) +
 # theme(legend.position= 'none') +
  #    theme(legend.position= 'none') +   
#viridis::scale_colour_viridis(discrete = TRUE)

 # print(p)
#  }

#scatplot(df_ref, Subsample, Abundance, facetby = Region_Name, 1,1,1)

 
#c) Scatterplots: Abundance vs percent sorted, by region - Reference sites
#(y=300 dotted line)
#df_ref %>%
#  filter(Subsample <= 100) %>%
#ggplot(mapping = aes(x = Subsample, y = Abundance, colour = Region_Name) ) +
##  geom_point(size=2) + 
#  geom_hline(yintercept = 300, linetype = "dotted") +
#facet_wrap(~Region_Name, labeller = labeller(Region_Name = label_wrap_gen(width = 25))) +
#      theme_bw() +
# theme(plot.title =  element_text(size = 10, hjust = .5, face = "bold")) +
#  ggtitle( "Percent Sorted vs Abundance \n (Reference/Potential Reference Sites) ") +
# ylab("Total Abundance (count) ") +
# xlab("Percent Sorted") + 
 # theme(legend.position= 'none') +
   # viridis::scale_colour_viridis(discrete = TRUE)



```

```{r, echo = FALSE, include = FALSE}

#sum(is.na(df_habitat$Status))

#sum(is.na(df_benthic$Status))

#df_both[is.na(df_both$Status),]


```