Electrofishing Assignment by Ditte Svinth and Aliki Fornier
2024-10-07
library(readxl) #For importing excel files.
library(tidyverse) #For all our data wrangling tools!## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.4 ✔ readr 2.1.5
## ✔ forcats 1.0.0 ✔ stringr 1.5.1
## ✔ ggplot2 3.5.1 ✔ tibble 3.2.1
## ✔ lubridate 1.9.3 ✔ tidyr 1.3.1
## ✔ purrr 1.0.2
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorslibrary(FSA) #for the removal function.## ## FSA v0.9.5. See citation('FSA') if used in publication.
## ## Run fishR() for related website and fishR('IFAR') for related book.library(ggplot2)File Directory
Use the following code to view the file directory in this project
list.files(path = ".", full.names = TRUE)## [1] "./assingment.R"
## [2] "./Bergvattensjöbäcken.csv"
## [3] "./Bergvattensjöbäcken.xlsx"
## [4] "./Electrofishin Assignment 2 and 3.R"
## [5] "./Electrofishing Assignment Markdown.R"
## [6] "./Electrofishing Assignment Markdown.Rmd"
## [7] "./Electrofishing_Lab.html"
## [8] "./Electrofishing-Assignment-Markdown.html"
## [9] "./Electrofishing-Assignment-Markdown.Rmd"
## [10] "./Electrofishing.R"
## [11] "./Electrofishing.Rproj"
## [12] "./Task 1.xlsx"
## [13] "./Task 10.xlsx"
## [14] "./Task 3.xlsx"
## [15] "./Task 4.xlsx"
## [16] "./Task 5.xlsx"
## [17] "./Task 6.xlsx"
## [18] "./Task 7.xlsx"
## [19] "./Task 8.xlsx"
## [20] "./Task 9.xlsx"Part 1: Analyse lengths of fish caught in Gällarbäcken (file “Task 4.xlsx”)
In this Part 1, we create 3x4 histograms with Site as columns and Year as rows. We begin by reading the file in question.
my_file <- "Task 4.xlsx"Read the fish lengths from “Task 4.xlsx”.
Table 1. Fish Lenghts for Brown Trout.
fishlengths <- read_excel(my_file, sheet="Fish lengths")
names(fishlengths)## [1] "LAN" "HFLODOMR" "Stream" "Fishing site" "XKOORLOK"
## [6] "YKOORLOK" "Fishing date" "Species" "Length (mm)"fishlengths <- fishlengths %>%
select(Site = `Fishing site`, Date = `Fishing date`, Species,
Length = `Length (mm)`) %>%
filter(Species == "Brown trout") %>%
mutate(Year = substr(Date,1,4))
fishlengths## # A tibble: 709 × 5
## Site Date Species Length Year
## <chr> <dbl> <chr> <dbl> <chr>
## 1 Clear-cut 19960809 Brown trout 49 1996
## 2 Clear-cut 19960809 Brown trout 50 1996
## 3 Clear-cut 19960809 Brown trout 50 1996
## 4 Clear-cut 19960809 Brown trout 50 1996
## 5 Clear-cut 19960809 Brown trout 51 1996
## 6 Clear-cut 19960809 Brown trout 52 1996
## 7 Clear-cut 19960809 Brown trout 52 1996
## 8 Clear-cut 19960809 Brown trout 53 1996
## 9 Clear-cut 19960809 Brown trout 53 1996
## 10 Clear-cut 19960809 Brown trout 53 1996
## # ℹ 699 more rows- Using select(), we keep columns of interest and rename some of them.
- Using filter(), we keep only Species == “Brown trout”.
- Using mutate(), we add a Column with Year.
Fish lengths histogram.
Now that the data is cleaned up, we can plot the fish lenghts data to get a histogram, by Site and Year.
ggplot(data = fishlengths, aes(Length, fill = Site)) +
geom_histogram(binwidth=5) +
theme_bw() + theme(legend.position = "none") +
facet_grid(Year ~ Site) +
ggtitle("Figure 1. Histogram of Fish Lengths") # Add the figure title
Fish lenghts histogram by site and year
- Using ggplot, we begin a process to create a histogram for fishlenghts.
- Using data, we define which dataset to use to create the plot.
- Using aes, we define what variable to plot.
- Using fill, we define what variable to fill the bar colors by.
- Using theme_bw(), we instruct to us a black on white theme for the plot.
- Using facet_grid, we instruct how to arrange the plots. In this case in Year by row, and Site by column.
Question 1: Analysis of the lenghts of fish in Gallarbacken (Task 4). Can we see any differences between sites and years.
The clear cut happened between 1996 and 1997, after the first removal survey was conducted in 1996. From Figure 1 above, we notice an decline in fish lengths, especially between 1998 and 1999 for the clear cut sections. On the other hand, we do notice a decline in fish lengths in 1998 for upstream and downstream sections, with an uptake in 1999. We would say that there is not enough data to really conclude that clear cut has an effect on fish lengths.
Part 2: Run FSA::removal() on all data in Task 4 file.
To proceed, we clean up the remaining of the Task 4 data and create a relation between the tabs within the spreadsheet. The tabs in question; Site_description and Fishing_results
Clean up Site Description Tab
Table 2. Site Description for Clearcutting Sites.
Site_description <- read_excel(my_file, sheet="Site description")
names(Site_description) ## [1] "Stream" "XKOORVDR"
## [3] "YKOORVDR" "Fishing site"
## [5] "Altitude" "XKOORLOK"
## [7] "YKOORLOK" "Fishing date"
## [9] "Fabricate" "Power"
## [11] "VOLT" "ampere"
## [13] "Pulse" "Removals"
## [15] "METOD" "Stream witdth"
## [17] "fished length" "Fished width"
## [19] "AREA (m2)" "max depth"
## [21] "mean depth" "bottom topo"
## [23] "SUBSTR1" "SUBSTR2"
## [25] "SUBSTR3" "water level"
## [27] "Water velocity" "Water temp"
## [29] "Air temp" "Amount of water veg"
## [31] "Bottom vegetation" "Local environment"
## [33] "Dom tree 1" "Dom tree 2"
## [35] "Shadowing (%)" "Wood in water (n)"
## [37] "Wood in water (n/100m2)" "Migration hinder"
## [39] "Population type trout" "Lime influenced"
## [41] "Municiple" "Map"
## [43] "Year" "Month"Site_description <- Site_description %>%
select(Stream, Site = `Fishing site`, Date = `Fishing date`, Fished_Area = `AREA (m2)`)
Site_description## # A tibble: 12 × 4
## Stream Site Date Fished_Area
## <chr> <chr> <dbl> <dbl>
## 1 Gällarbäcken Clear-cut 19960809 524
## 2 Gällarbäcken Clear-cut 19970819 471
## 3 Gällarbäcken Clear-cut 19980816 535
## 4 Gällarbäcken Clear-cut 19990813 374
## 5 Gällarbäcken Down-stream clear-cut 19960810 466.
## 6 Gällarbäcken Down-stream clear-cut 19970820 477
## 7 Gällarbäcken Down-stream clear-cut 19980816 519.
## 8 Gällarbäcken Down-stream clear-cut 19990813 392.
## 9 Gällarbäcken Up-stream clear-cut 19960810 464
## 10 Gällarbäcken Up-stream clear-cut 19970820 475
## 11 Gällarbäcken Up-stream clear-cut 19980816 540
## 12 Gällarbäcken Up-stream clear-cut 19990813 410- Using select, we keep and rename relevant columns.
- Using %>%, we shift from one task to another in a strategic order.
Clean up Fishing Results Tab
fishingresults <- read_excel(my_file, sheet="Fishing result")
names(fishingresults)## [1] "xkoorlok" "ykoorlok" "Stream" "Fishing site" "Fishing date"
## [6] "Species" "year class" "Remove1" "Remove2" "Remove3"fishingresults <- fishingresults %>%
select(Stream, Site = `Fishing site`, Date = `Fishing date`, Species, Year_class = `year class`,
Remove1, Remove2, Remove3) %>%
filter(Species == "Brown trout") %>%
mutate(Year = substr(Date,1,4))- Using select, we keep and rename relevant columns.
- Using filter, we keep data for “Brown trout”.
- Using mutate, we create a new column for Year, from the Date column.
- Using %>%, we shift from one task to another in a strategic order.
Combine tabs
Next we combine columns from Fishing results tab to the Site description tab.
Table 3. Fishing Results and Site Description.
Your_task <- Site_description %>%
left_join(
fishingresults,
by = c("Stream", "Site", "Date"))
Your_task## # A tibble: 24 × 10
## Stream Site Date Fished_Area Species Year_class Remove1 Remove2 Remove3
## <chr> <chr> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <dbl>
## 1 Gällarbä… Clea… 2.00e7 524 Brown … >0+ 21 15 9
## 2 Gällarbä… Clea… 2.00e7 524 Brown … 0+ 12 8 3
## 3 Gällarbä… Clea… 2.00e7 471 Brown … >0+ 35 11 9
## 4 Gällarbä… Clea… 2.00e7 471 Brown … 0+ 23 22 6
## 5 Gällarbä… Clea… 2.00e7 535 Brown … >0+ 5 3 2
## 6 Gällarbä… Clea… 2.00e7 535 Brown … 0+ 2 2 0
## 7 Gällarbä… Clea… 2.00e7 374 Brown … >0+ 7 3 2
## 8 Gällarbä… Clea… 2.00e7 374 Brown … 0+ 3 0 2
## 9 Gällarbä… Down… 2.00e7 466. Brown … >0+ 25 19 11
## 10 Gällarbä… Down… 2.00e7 466. Brown … 0+ 20 14 8
## # ℹ 14 more rows
## # ℹ 1 more variable: Year <chr>- Using left_join, we create new columns with values from fishingresults, to the right side of the site_description data frame.
- Using by, we align Stream, Site and Date values.
FSA::removal function.
the removal fucntion can be applied to our data to estimates the population of fish using the removal function.
removal_result <- Your_task %>%
rowwise() %>%
mutate(res = list(removal(c(Remove1, Remove2, Remove3), just.ests = TRUE))) %>%
unnest_wider(col = res) - Using removal function we get population estimate and catch probability.
- Using rowise, we process each row from “Your_task” data frame one at a time.
- Using mutate, we create a new column for each new removal value.
- Using unnest_wider, we unpack the combined variables within the res-vector column and creates new columns with the data seperated.
- Using %>%, we shift from one task to another in a strategic order.
Density
We add density (fish/100m^2) and its CI to the removal_result data frame.
removal_result <- removal_result %>%
mutate(Dens = 100 * No / Fished_Area,
Dens_LCI = 100 * No.LCI / Fished_Area,
Dens_UCI = 100 * No.UCI / Fished_Area)- Using mutate, we create a new column for the density value and their CI limits.
Run Removal one row at a time.
We learned how to run the removal function, on one row at a time, using the following process.
one_site <- Your_task %>%
filter(Site == 'Clear-cut', Year_class == ">0+", Year == "1996")
est <- removal(catch = c(one_site$Remove1, one_site$Remove2, one_site$Remove3),
just.ests = TRUE)
fish_dens <- 100 * est["No"] / one_site$Fished_Area- Using Filter, we select certain criteria, which in this case, separates the first row from the rest of the data.
- Using removal function we get population estimate and catch probability, for the first row.
- Using the plain formula in fish_dens, we calculate the fish population density, per 100 sq.m.
Probability of Removal Proportion
Using the PRP we can expresses the average width of the confidence interval as a percentage of the estimate.
PRP <- function(N, LCI, UCI) {
return(50 * (UCI - LCI)/ N)
}- Using function (N, LCI, UCI), we define the inputs.
- Using curly braces {}, we contain the logic.
- Using return(), we ensure that the result of the formula of inputs, is returned when the function is run.
- Using 50× multiplier, we are applying a scaling factor to the result of the ratio. The factor of 50 is used to ensure that the PRP gives a result that is around 100% or less for most reasonable estimates and confidence interval widths. The 50x multiplier makes the interpretation of the results more logical.
Once we have the PRP function defined, we can apply it to the removal_results data frame.
removal_result <- removal_result %>%
mutate(p_prp = PRP(p, p.LCI, p.UCI)) %>%
mutate(fish_caught = Remove1 + Remove2 + Remove3)- Using mutate, we create a new column called p_prp to house the results of the PRP fuction.
- Using mutate, we create a new column called fish_caught to house the sum of removals.
To present the data in a more visual way, we plot them in scatter plot with arrow bars.
Catchability Plot
We run a catchability plot, which tells us the probability of catching an individual during our sampling efforts.
ggplot(data = removal_result, aes(Year, p, colour=Year)) +
geom_point(size=4) +
geom_errorbar(aes(ymin = p.LCI, ymax = p.UCI,colour=Year), position = "dodge", width = 0.25) +
theme_bw() +
theme(axis.text.x = element_text(size=14),
axis.text.y = element_text(size=14),
strip.text = element_text(size=14)) +
facet_grid(rows = vars(Year_class), cols = vars(Site)) +
ggtitle("Figure 2a. Catchability for Sites and Years")
- Using ggplot, we plot the p values (catchebiloty) with their corresponding CI levels
- Using position_dodge, we align the arrow bars centrally to the p_value (dot)
- Using themw_bw, we define the coloring of the scatter plots
- Using theme, we define the axis text and size.
- Using faced_grid, we combine the scatter plots and order them according to year_class (row) and sites (columns).
Density Plot
We run a density plot to estimate the density of individuals in a survey area.
ggplot(data = removal_result, aes(Year, Dens, colour = Year)) +
geom_point(size=4) +
geom_errorbar(aes(ymin = Dens_LCI, ymax = Dens_UCI, colour=Year),
position = "dodge", width = 0.25) +
theme_bw() +
theme(axis.text.x = element_text(size=14, angle = -90),
axis.text.y = element_text(size=14),
strip.text = element_text(size=14)) +
facet_grid(rows = vars(Year_class), cols = vars(Site))+
ggtitle("Figure 2b. Density for Sites and Years")
- Using ggplot, we plot the Dens values (density) with their corresponding CI levels
- Using position_dodge, we align the arrow bars centrally to the p_value (dot)
- Using themw_bw, we define the coloring of the scatter plots
- Using theme, we define the axis text and size.
- Using faced_grid, we combine the scatter plots and order them according to year_class (row) and sites (columns).
Run FSA::removal on data in all Task files
To recreate this process for all Task files, we begin by creating a function called slurp_xlsx to read and combine all the Task files together.
slurp_xlsx <- function(fnames, sheet = 1) {
res <- lapply(fnames, readxl::read_xlsx, sheet = sheet)
res <- dplyr::bind_rows(res)
return(res)
}- Using function (fnames, sheet = 1)), we define the inputs.
- Using fnames, we define unique file names for the function to be applied to each.
- Using sheet = 1, we create a default instruction to use the first sheet in a file, unless specified.
- Using curly braces {}, we contain the logic.
- Using lapply, we avoid
- Using “for loops”, and we apply the function to each file name in fnames.
- Using readxl::read_xlsx is a function, to read the Excel files.
- Using dplyr::bind_rows we combine all the rows from the relevant file sheets into one.
- Using return(), we ensure that the result of the formula of inputs, is returned when the function is run.
Fill a list with all files
Use the following code to combine the relevant files into a vector.
all_files <- c("Task 1.xlsx", "Task 3.xlsx", "Task 10.xlsx","Task 4.xlsx", "Task 5.xlsx", "Task 6.xlsx", "Task 7.xlsx", "Task 8.xlsx", "Task 9.xlsx")Since the files used, are stored in a folder containing other unrelevant files, the list.files()) function can not be used.
Data merging and cleanup
We use the following code with the new slurp function to combine all site description tasks from each Task file.
Site_description <- slurp_xlsx(all_files, sheet="Site description")After that we clean the new sheet, selection and renaming the relevant columns.
Site_description <- Site_description %>%
select(Stream, Site = `Fishing site`, Date = `Fishing date`, Fished_Area = `AREA (m2)`)
unique(Site_description$Stream)## [1] "Bergvattensjöbäcken" "Gulån" "Östra Bjurbäcken"
## [4] "Gällarbäcken" "Mörttjärnsbäcken" "Rötjärnsbäcken"
## [7] "Svartbäcken" "Vallsbäcken" "Västra Dalkarlsån"We rerun the above process for the fishing_results sheets.
fishingresults <- slurp_xlsx(all_files, sheet="Fishing result")We clean fishing results with a filter() to save only “Brown trout”, keep only relevant columns. We also add a Year column.
fishingresults <- fishingresults %>%
select(Stream, Site = `Fishing site`, Date = `Fishing date`, Species, Year_class = `year class`,
Remove1, Remove2, Remove3) %>%
filter(Species == "Brown trout") %>%
mutate(Year = substr(Date,1,4))
unique(fishingresults$Stream) #View unique streams for the dataset## [1] "Bergvattensjöbäcken" "Gulån" "Östra Bjurbäcken"
## [4] "Gällarbäcken" "Mörttjärnsbäcken" "Rötjärnsbäcken"
## [7] "Svartbäcken" "Vallsbäcken" "Västra Dalkarlsån"Explanation for the content of this code, review the code from part 1 of the assignment
Combine tabs to continue analysis
To get estimates we need to know the fished area available in Site_description. Therefore combine the two data frames into one using left_join()
Table 4. Site Description and Fishing Results for all Tasks
total_data <- fishingresults %>%
left_join(Site_description,
by = c("Stream", "Site", "Date"))
total_data## # A tibble: 216 × 10
## Stream Site Date Species Year_class Remove1 Remove2 Remove3 Year
## <chr> <chr> <dbl> <chr> <chr> <dbl> <dbl> <dbl> <chr>
## 1 Bergvattensjöb… Clea… 2.00e7 Brown … >0+ 22 7 3 1996
## 2 Bergvattensjöb… Clea… 2.00e7 Brown … >0+ 22 15 5 1997
## 3 Bergvattensjöb… Clea… 2.00e7 Brown … >0+ 4 7 3 1998
## 4 Bergvattensjöb… Clea… 2.00e7 Brown … >0+ 25 13 8 1999
## 5 Bergvattensjöb… Down… 2.00e7 Brown … >0+ 11 4 2 1996
## 6 Bergvattensjöb… Down… 2.00e7 Brown … >0+ 21 11 2 1997
## 7 Bergvattensjöb… Down… 2.00e7 Brown … >0+ 8 4 0 1998
## 8 Bergvattensjöb… Down… 2.00e7 Brown … >0+ 12 4 7 1999
## 9 Bergvattensjöb… Up-s… 2.00e7 Brown … >0+ 15 7 4 1996
## 10 Bergvattensjöb… Up-s… 2.00e7 Brown … >0+ 24 10 3 1997
## # ℹ 206 more rows
## # ℹ 1 more variable: Fished_Area <dbl>Population estimates
Using the removal function we can now calculate population estimates for the entire data set.
removal_result <- total_data %>%
rowwise() %>%
mutate(res = list(removal(c(Remove1, Remove2, Remove3), just.ests = TRUE))) %>%
unnest_wider(col = res) - Using removal function we get population estimate and catch probability.
- Using rowise, we process each row from “Your_task” data frame one at a time.
- Using mutate, we create a new column for each new removal value.
- Using just.est = TRUE, we the function to only return the estimate.
- Using unnest_wider, we unpack the combined variables within the res-vector column and creates new columns with the data seperated.
- Using %>%, we shift from one task to another in a strategic order.
Density estimates
We use the density formula (fish/100m^2) in combination with fished area and population estimate to get a density estimate.
removal_result <- removal_result %>%
mutate(Dens = 100 * No / Fished_Area,
Dens_LCI = 100 * No.LCI / Fished_Area,
Dens_UCI = 100 * No.UCI / Fished_Area)- Using mutate, we create a new column for the density value and their CI limits
Probability of Removal Proportion
Again, using the PRP we can expresses the average width of the confidence interval as a percentage of the estimate, for the entire data set this time.
PRP <- function(N, LCI, UCI) {
return(50 * (UCI - LCI)/ N)
}Explanation for the content of this code, review the code from part 2 of the assignment
removal_result <- removal_result %>%
mutate(p_prp = PRP(p, p.LCI, p.UCI)) %>%
mutate(fish_caught = Remove1 + Remove2 + Remove3)- Using mutate, we create a new column called p_prp to house the results of the PRP fuction.
- Using mutate, we create a new column called fish_caught to house the sum of removals.
The complete removal result data frame can be viewed in the environment tab.
We now have a data frame with estimates from 9 different streams, each fished at three different sites and four years.
unique(removal_result$Stream) # Show streams## [1] "Bergvattensjöbäcken" "Gulån" "Östra Bjurbäcken"
## [4] "Gällarbäcken" "Mörttjärnsbäcken" "Rötjärnsbäcken"
## [7] "Svartbäcken" "Vallsbäcken" "Västra Dalkarlsån"unique(removal_result$Site) # Show site types## [1] "Clear-cut" "Down-stream clear-cut" "Up-stream clear-cut"unique(removal_result$Year) # Show years## [1] "1996" "1997" "1998" "1999"Part 3: Summarize Mean Density
We create a summary with the mean density for all fishing in this study grouped by Site, Year and treatment. The overall summary is named total_summary.
total_summary <- removal_result %>%
group_by(Site, Year, Year_class) %>%
summarise(Dens_mean = mean(Dens),
Dens_LCI = Dens_mean - FSA::se(Dens),
Dens_UCI = Dens_mean + FSA::se(Dens),
.groups = "drop")- Using group_by, we sort the data in the mentioned order, and is used in combination with the summarise function.
- Using summarise, we simplify the dataset by getting the density mean values, for each year and year_class.
Plot to display density development over the years
We create a figure that shows the development of the population at each treatment (Site) and separate plots for the two Year classes.
summary_plot <- ggplot(data = total_summary, mapping = aes(x = as.numeric(Year), y = Dens_mean)) +
geom_point(aes(colour = Year), size = 4) + # Add 'aes(colour = Year)' here
geom_errorbar(aes(ymin = Dens_LCI, ymax = Dens_UCI, colour = Year),
position = "dodge", width = 0.25) +
theme_bw() +
theme(axis.text.x = element_text(size = 14, angle = -90),
axis.text.y = element_text(size = 14),
strip.text = element_text(size = 14)) +
facet_grid(rows = vars(Year_class), cols = vars(Site)) +
ggtitle("Figure 3. Mean Density at Sites and Years")
summary_plot # Show the plot
- Using ggplot, we plot the mean Dens values (density) with their corresponding CI levels
- Using position_dodge, we align the arrow bars centrally to the mean value (dot)
- Using themw_bw, we define the coloring of the scatter plots
- Using theme, we define the axis text and size.
- Using faced_grid, we combine the scatter plots and order them according to year_class (row) and sites (columns).
We saved the figure with ggsave(“filename.png”, summary_plot) and interactively in Rstudio.
Question 2: Create a summary with the mean density for all fishing in the study, grouped by site, year and treatment. Can we find and show any effects of the clearcutting?
Here we analyse the results of Figure 3.
For the older trout (>0+), we notice a steady decline in population density at the clear cut site, from before the clear cut event, to 3 years after. In contrast, for the upstream and downstream sites, the population density can be seen rising immediately after the clear cut event, then declining the second year, with an uptake at the third year.
For the young of the year (0+), we see an uptake in population density immediately after the clear cut event, then declining the second year, with an uptake at the third year, for clear cut and upstream sites. But we can see an overall increase in population density at the downstream site.
Considering the scale of the removal survey, and the confidence interval limits displayed in Figure 2, and the randomness of the data, we conclude that this survey is inconclusive to determine the effect clear cutting had on Brown Trout. Though if did have to make a conclusion, we would say that clear cutting has a negative effect on adults, especially at clear cut sites, but a high likely hood of density bouncing back after a few years. And clear cutting has a positive effect on young, especially at down stream sites.
Part 4: Determine Data quality and Precision
One of the assumptions of the removal method is that the number of
caught individuals get lower for each pass. The logical expression
(Remove1 > Remove2 & Remove2 > Remove3) will be
TRUE for all rows in removal_result that fulfill this criteria.
We use the removal_result dataset to find out how many times the result don’t fulfill the criteria.
nrow(removal_result)## [1] 216good_rows <- removal_result %>%
dplyr::filter(Remove1 > Remove2 & Remove2 > Remove3)
nrow(good_rows)## [1] 130- Using nrow, we can get the total number of rows in the dataset.
- Using dplyr::filter(), we filter rows that satisfy the condition Remove1 > Remove2 & Remove2 > Remove3.
- Using nrow again, we display how many rows fit the assumption.
Bonus Question: Examine removal_result to find out how many times the result don’t fulfill the criteria. Suggest how to deal with the problem.
To get a more precise data set, we could consider removing the rows that don’t fit the assumption. Prior to the survey, though, we could look at the removal method techniques and site selection to secure better results, without being bias. The assumption is that each removal will be less than the last during a survey, data which doesn’t fulfill that assumption, makes the data set uncertain and risks skewing the results.
Probability for Removal Proportion
Finally, we create figures that show how the total catch (fish_caught) affects PRP of catchabillity (p_prp in removal_result).
We decided to separate the PRP analysis, according to Year Class.
PRP plot for 0+ and >0+ Year Class
# Filter the dataset for year_class "0+"
removal_result_0 <- removal_result %>% filter(Year_class == "0+")
# Create the plot for year_class "0+"
PRP_plot_0 <- ggplot(data = removal_result_0, aes(x = fish_caught, y = p_prp)) +
geom_point(size = 3) +
geom_smooth(method = "lm", se = TRUE, color = "blue", size = 1.2, fullrange = TRUE) +
theme_bw() +
labs(x = "Total number of marked animals in previous catches",
y = "Catches of unmarked animals (PRP)",
title = "Figure 4a. Year Class 0+") +
theme(axis.title.x = element_text(size = 14),
axis.title.y = element_text(size = 14),
axis.text.x = element_text(size = 12),
axis.text.y = element_text(size = 12))## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.# Print the plot for year_class "0+"
print(PRP_plot_0)## `geom_smooth()` using formula = 'y ~ x'## Warning: Removed 27 rows containing non-finite outside the scale range
## (`stat_smooth()`).## Warning: Removed 27 rows containing missing values or values outside the scale range
## (`geom_point()`).
# Filter the dataset for year_class ">0+"
removal_result_above_0 <- removal_result %>% filter(Year_class == ">0+")
# Create the plot for year_class ">0+"
PRP_plot_above_0 <- ggplot(data = removal_result_above_0, aes(x = fish_caught, y = p_prp)) +
geom_point(size = 3) +
geom_smooth(method = "lm", se = TRUE, color = "blue", size = 1.2, fullrange = TRUE) +
theme_bw() +
labs(x = "Total number of marked animals in previous catches",
y = "Catches of unmarked animals (PRP)",
title = "Figure 4b. Year Class >0+") +
theme(axis.title.x = element_text(size = 14),
axis.title.y = element_text(size = 14),
axis.text.x = element_text(size = 12),
axis.text.y = element_text(size = 12))
# Print the plot for year_class ">0+"
print(PRP_plot_above_0)## `geom_smooth()` using formula = 'y ~ x'## Warning: Removed 7 rows containing non-finite outside the scale range
## (`stat_smooth()`).## Warning: Removed 7 rows containing missing values or values outside the scale range
## (`geom_point()`).
Question 3: Analyse the quality of the data and the precision of our estimates. Describe weakness in the data and make some suggestions on how to make improvements.
For both year classes we have a decline regression line, which indicates that the assumption that consecutive catches are lower than the last, is true. In year class 0+, the regression line is not as steep as the one for year class >0+, that could be due to a smaller variation between each removal, or indication that some of the removals had higher catches than the previous ones.
A more thorough clean up of the data set could help with that, such as removing observations where catches were higher than the previous ones. It would also improve the visibility of the data set to create a PRP plot for each treatment type within each year class.
We can save out figure with ggsave(“filename.png”, PRP_plot) or interactively in Rstudio
Bonus question: Count how many rows have p_prp above and below 50 %.
98 rows are below 50%.
84 rows are above 50%.
34 rows are NA due to no catches on some of the removals.