The Cat’s Meow: Tracking Outdoor Cat Movement over Time

Final Project / MATH241 / Spring 2024

Author

Team Cat, aka Sydney Stitt

So, What is a Cat?

This is cat, his name is Moony. He is the most spoiled, silly cat out there!

Data

  • Two datasets

    • cat_us: GPS coordinates & timestamps of 140+ cats from the United States from 2013-2016

    • cat_us_reference: metadata on individual cat lifestyles (sex, age, diet, hunting, number of cat siblings) and data collection metadata

  • Data collected by volunteer cat owners from 12 geographically diverse US states

  • Initially used to analyze the ecological impacts of outdoor cats

Research Question

  1. Do younger cats tend to travel longer distances than older cats?

  2. Do cats travel in patterns or “seemingly” randomly (based on environmental stimulus or reactions)?

Data Wrangling

659,770 observations into 62,543 observations! How did I do it?

  1. Remove cats with NA metadata observations (not super useful without metadata) or locations outside the US

  2. Remove unnecessary variables and tidy up observations (time stamps and comments)

  3. Remove outliers (manually and algorithm marked)

This is Getting Messy! Calculating Distance!

  1. Create a function that calculates the distance between two sets of coordinates into a new variable using the Haversine formula

  2. Create individual datasets for each cat

  3. Apply Haversine function to calculate distances for each cat

  4. Take the averages of distances per day, per 3 minute observation cycle, per cat, and per age group for comparison

Coding…

#Tidying up the date and time variables in both dataframes
cat_movement <- cat_movement %>% 
  mutate(Time = format(as.POSIXct(timestamp), 
                       format = "%H:%M:%S"),
         Date = as.Date(timestamp)) %>%
  filter(visible) %>%
  subset(select = -timestamp)
cat_reference <- cat_reference %>%
  mutate(`Date-Deploy-On` = as.Date(`deploy-on-date`),
         `Time-Deploy-On` = format(as.POSIXct(`deploy-on-date`), 
                                   format = "%H:%M:%S"),
         `Date-Deploy-Off` = as.Date(`deploy-off-date`),
         `Time-Deploy-Off` = format(as.POSIXct(`deploy-off-date`), 
                                    format = "%H:%M:%S")) %>%
  subset(select = -`deploy-off-date`) %>%
  subset(select = -`deploy-on-date`)
#ordering data alphabetically by animal id
cat_reference <- cat_reference[order(cat_reference$`animal-id`), ]

Haversine Distance Function

# Custom distance function using Haversine formula to convert coordinates to distance
haversine_distance <- function(lon1, lat1, lon2, lat2) {
  lon1_rad <- lon1 * pi / 180
  lat1_rad <- lat1 * pi / 180
  lon2_rad <- lon2 * pi / 180
  lat2_rad <- lat2 * pi / 180
  # Earth radius in kilometers
  R <- 6371
  # Haversine formula
  dlon <- lon2_rad - lon1_rad
  dlat <- lat2_rad - lat1_rad
  a <- sin(dlat/2)^2 + cos(lat1_rad) * cos(lat2_rad) * sin(dlon/2)^2
  c <- 2 * atan2(sqrt(a), sqrt(1-a))
  distance <- R * c
  return(distance)
}

# Function to calculate distances and convert to meters
calculate_distances <- function(data) {
  distance <- haversine_distance(data$`location-long`[-nrow(data)], data$`location-lat`[-nrow(data)], 
                                  data$`location-long`[-1], data$`location-lat`[-1])
  distance <- c(NA, distance)  # Add NA to match the size of the original data frame
  data <- mutate(data, distance = distance)
  return(data)
}

More Coding! Wrangling and Tidying Data

#convert distance from Km to m
adjusted_cat_distance <- adjusted_cat_distance %>%
  mutate(distance = distance*1000)
#Disable scientific notation
options(scipen = 999)
#Creating variables for average distances per 3 minute cycle and per day
summaries_by_day <- adjusted_cat_distance %>%
  filter(distance <= 1000) %>%
  group_by(`individual-local-identifier`, Date) %>%
  summarize(mean_distance = mean(distance, 
                                 na.rm = TRUE),
            sum_distance = sum(distance, 
                               na.rm = TRUE))

#Creating variables for averages for each cat
summaries_by_cat <- summaries_by_day %>%
  group_by(`individual-local-identifier`) %>%
  summarize(mean_distance_obs = mean(mean_distance),
            avg_distance_day = mean(sum_distance)) 

cat_reference <- cat_reference %>%
  filter(!is.na(`animal-life-stage`)) %>%
  filter(`animal-id` != "Banjo2") %>%
  filter(`animal-id` != "Boris2") %>%
  filter(`animal-id` != "Rocky") %>%
  filter(`animal-id` != "Striper2") %>%
  filter(`animal-id` != "Ziggy") 

#Binding the reference cat dataset with their average distances per cycle and per day, removing any cat with no age listed
merged_summaries_by_cat <- cbind(summaries_by_cat, 
                                 cat_reference) 

#Tidying up the life stage variable by removing "years" in each observations
merged_summaries_by_cat$`animal-life-stage` <- gsub(" years", 
                                                    "", 
                                                    merged_summaries_by_cat$`animal-life-stage`)
merged_summaries_by_cat$`animal-life-stage` <- gsub(" year", 
                                                    "", 
                                                    merged_summaries_by_cat$`animal-life-stage`)

#Tidying up the animal comments variable by separating it into two variables for hunting and prey per month
merged_summaries_by_cat <- cbind(merged_summaries_by_cat, 
                                 do.call(rbind, 
                                         strsplit(merged_summaries_by_cat$`animal-comments`, 
                                                  "; ", 
                                                  fixed = TRUE)))
merged_summaries_by_cat <- merged_summaries_by_cat %>%
  subset(select = -`3`) %>%
  rename(`Hunt` = `1`,
         `prey_p_month` = `2`)
merged_summaries_by_cat$hunt <- ifelse(grepl("Yes", 
                                             merged_summaries_by_cat$Hunt), 
                                       TRUE, FALSE)
merged_summaries_by_cat$prey_p_month <- as.numeric(gsub("prey_p_month: ", 
                                                        "", 
                                                        merged_summaries_by_cat$prey_p_month))

#Tidying up the manipulation comments variable by separating it into two variables for hours spent indoors and number of cats in the household 
merged_summaries_by_cat <- cbind(merged_summaries_by_cat, do.call(rbind, strsplit(trimws(merged_summaries_by_cat$`manipulation-comments`), ";", fixed = TRUE)))
merged_summaries_by_cat <- merged_summaries_by_cat %>%
  subset(select = -`3`) %>%
  rename(`hrs_indoors` = `1`,
         `n_cats_household` = `2`)
merged_summaries_by_cat$hrs_indoors <- as.numeric(gsub("hrs_indoors: ", 
                                                       "", 
                                                       merged_summaries_by_cat$hrs_indoors))
merged_summaries_by_cat$n_cats_household <- as.numeric(gsub("n_cats: ", 
                                                            "", 
                                                            merged_summaries_by_cat$n_cats_household))
merged_summaries_by_cat <- merged_summaries_by_cat %>%
  mutate(`animal-life-stage` = as.numeric(`animal-life-stage`))
#Getting rid of unnecessary variables and observations outside the United States
merged_summaries_by_cat <- merged_summaries_by_cat %>%
  subset(select = -`animal-comments`) %>%
  subset(select = -`manipulation-comments`) %>%
  subset(select = -`Hunt`) %>%
  subset(select = -`animal-taxon`) %>%
  subset(select = -`attachment-type`) %>%
  subset(select = -`data-processing-software`) %>%
  subset(select = -`deployment-end-type`) %>%
  subset(select = -`duty-cycle`) %>%
  subset(select = -`manipulation-type`) %>%
  subset(select = -`tag-manufacturer-name`) %>%
  subset(select = -`tag-mass`) %>%
  subset(select = -`tag-model`) %>%
  subset(select = -`tag-readout-method`) %>%
  filter(`study-site` != "Denmark" & `study-site` != "Newfoundland")

Data Visualizations

  • Will eventually make a Shiny app to have user customize which cat and which day they want to visualize

  • For now, these are the basic visualizations for what I want the app to produce

Map 1

#Function
plot_cat_movement <- function(adjusted_cat_distance, cat_identifier, date) {
  
  motorcat_cat <- adjusted_cat_distance %>%
    filter(`individual-local-identifier` == cat_identifier) %>%
    filter(Date == date) %>%
    mutate(Time = as.POSIXct(Time, format = "%H:%M:%S", tz = "UTC")) %>%  # Convert to POSIXct
    arrange(Time)
  
  # Generate title dynamically
  title <- paste("Location Changes Over Time for", cat_identifier, "on", date)
  
  ggplot(motorcat_cat, aes(x = `location-long`, y = `location-lat`, group = 1, color = Time)) +
    geom_path() + 
    geom_point() +
    scale_color_viridis_c() +
    geom_text(aes(label = format(Time, "%H:%M:%S")), hjust = 1.2, vjust = 0) +  # Add time stamps
    labs(x = "Longitude", y = "Latitude", title = title)
}

# Usage example
plot_cat_movement(adjusted_cat_distance, "Motorcat", "2013-08-14")

Map 2

Map 3

Graph 1

age_summaries <- merged_summaries_by_cat %>%
  group_by(`animal-life-stage`) %>%
  summarize(avg_dist_day_by_age = mean(avg_distance_day),
            avg_dist_obs_by_age = mean(mean_distance_obs))
ggplot(age_summaries, aes(x = `animal-life-stage`, y = avg_dist_day_by_age)) +
  geom_col(color = "#45105b", fill = "#9a94bc") +
  labs(x = "Age in Years", y = "Average Distance in m", title = "Average Distance Traveled By Cats Each Day, \nGrouped by Age")

Graph 2

age_summaries <- merged_summaries_by_cat %>%
  group_by(`animal-life-stage`) %>%
  summarize(avg_dist_day_by_age = mean(avg_distance_day),
            avg_dist_obs_by_age = mean(mean_distance_obs))
ggplot(age_summaries, aes(x = `animal-life-stage`, y = avg_dist_obs_by_age)) +
  geom_col(color = "#45105b", fill = "#9a94bc") +
  labs(x = "Age in Years", y = "Average Distance in m", title = "Average Distance Traveled By Cats per 3-Min Cycle, \nGrouped by Age")

Data Analysis

  • Age is a strong factor in a cat’s ability to travel distances per day

    • Cats aged 10-15 in this dataset travel further than cats younger than 10

  • Age is not an strong factor in a cat’s ability to travel distances per 3 minute cycle

    • Younger cats travel, on average, slightly further distances

Data Analysis

  • Cats, for the most part, stay clustered in one area and often return to one specific “hub”

    • Reasonable assumption that this is their home, and they are emotionally and physically attached to their home and their owner

Caveats/Limitations

  • Timestamp labels on travel maps can be hard to read, hopefully fixed with Shiny app

  • Cats don’t travel in straight lines

    • Observations calculated every 3 minutes, so the distance variable isn’t completely accurate to the total distance traveled
    • 3 minutes is a relatively good interval, but cats don’t move in a 2D linear fashion and often move back and forth and up and down

  • Not all cats have the same number of days observed

Further Research

  • Observational data cannot be used to draw causal conclusions

    • So not representative of all cat behaviors!

  • Future experimental studies with more robust design to see whether age affects a cat’s likelihood of moving further distances

  • Future psychological studies to see what impacts a cat’s movement decisions

Citations

  • Kays R, Dunn RR, Parsons AW, Mcdonald B, Perkins T, Powers S, Shell L, McDonald JL, Cole H, Kikillus H, Woods L, Tindle H, Roetman P (2020) The small home ranges and large local ecological impacts of pet cats. Animal Conservation. doi:10.1111/acv.12563

Thank you for listening! Any questions?

This is also a cat! Her name is Cali.