Does the amount of time domestic cats in the United Kingdom spend indoors affect the number of prey animals they catch?
Audrey Allen and Arya Singh
2025-04-20
Introduction
This is a Tidy Tuesday project using data from the weekly social data project. This data tract 100 domesticated cats in the United Kingdom. It documents behavior such as hunting patterns, times indoors, age of cat, and several other characteristics. Using this data, we are looking to compare the hunting success of the cat and see if their age influences their success as hunters.
Domestic cats are natural hunters, and their tendency to catch prey can vary widely between individuals. One factor that may influence a cat’s hunting behavior is age. As cats grow older, their physical abilities, energy levels, and motivation to hunt can change. Younger cats, for example, may be more active and agile, making them more effective hunters. In contrast, older cats may hunt less frequently due to decreased stamina or health-related limitations.
Understanding how age affects hunting behavior in cats can provide valuable insights into feline research. It also helps explain some of the variation observed in prey-catching behavior across the cat population. In this study, we investigated the relationship between a cat’s age and the number of prey it catches, using data collected from domestic cats in the United Kingdom. Our goal is to better understand how this biological factor influences one of the most instinctive behaviors in domestic cats. The response variable is the hunting success of the cat and the explanatory variable is the age of the cat.
We were interested in the topic because we both really like cats and live in Philadelphia where there is a large amount of street cats. Although we looked at the data of domestic cats this still was of interest to us.
Set Up and Loading Data
knitr::opts_chunk$set(
echo = TRUE,
message = TRUE,
warning = TRUE
)
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, uniontuesdata <- tidytuesdayR::tt_load('2023-01-31')## ---- Compiling #TidyTuesday Information for 2023-01-31 ----## --- There are 2 files available ---
##
##
## ── Downloading files ───────────────────────────────────────────────────────────
##
## 1 of 2: "cats_uk.csv"
## 2 of 2: "cats_uk_reference.csv"tuesdata <- tidytuesdayR::tt_load(2023, week = 5)## ---- Compiling #TidyTuesday Information for 2023-01-31 ----
## --- There are 2 files available ---
##
##
## ── Downloading files ───────────────────────────────────────────────────────────
##
## 1 of 2: "cats_uk.csv"
## 2 of 2: "cats_uk_reference.csv"cats_uk <- tuesdata$cats_uk
cats_uk_reference <- tuesdata$cats_uk_reference
cats_uk_reference## # A tibble: 101 × 16
## tag_id animal_id animal_taxon deploy_on_date deploy_off_date hunt
## <chr> <chr> <chr> <dttm> <dttm> <lgl>
## 1 Tommy-T… Tommy Felis catus 2017-06-03 01:02:09 2017-06-10 02:10:52 TRUE
## 2 Athena Athena Felis catus 2017-06-24 01:02:13 2017-06-30 23:59:32 TRUE
## 3 Ares Ares Felis catus 2017-06-24 01:03:57 2017-06-30 23:58:01 NA
## 4 Lola Lola Felis catus 2017-06-24 01:18:49 2017-06-30 09:04:40 TRUE
## 5 Maverick Maverick Felis catus 2017-06-25 01:04:35 2017-07-03 09:10:07 TRUE
## 6 Coco Coco Felis catus 2017-06-28 01:02:22 2017-07-05 06:50:04 TRUE
## 7 Charlie Charlie Felis catus 2017-06-28 01:03:33 2017-07-04 07:08:31 TRUE
## 8 Jago Jago Felis catus 2017-06-28 04:10:59 2017-07-04 02:50:05 TRUE
## 9 Morpheu… Morpheus Felis catus 2017-07-01 01:02:18 2017-07-09 06:39:05 TRUE
## 10 Nettle-… Nettle Felis catus 2017-07-01 01:05:50 2017-07-08 00:05:12 TRUE
## # ℹ 91 more rows
## # ℹ 10 more variables: prey_p_month <dbl>, animal_reproductive_condition <chr>,
## # animal_sex <chr>, hrs_indoors <dbl>, n_cats <dbl>, food_dry <lgl>,
## # food_wet <lgl>, food_other <lgl>, study_site <chr>, age_years <dbl>cats_uk <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/main/data/2023/2023-01-31/cats_uk.csv')## Rows: 18215 Columns: 11
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (2): tag_id, study_name
## dbl (5): event_id, location_long, location_lat, ground_speed, height_above_...
## lgl (3): visible, algorithm_marked_outlier, manually_marked_outlier
## dttm (1): timestamp
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.cats_uk_reference <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/main/data/2023/2023-01-31/cats_uk_reference.csv')## Rows: 101 Columns: 16
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (6): tag_id, animal_id, animal_taxon, animal_reproductive_condition, an...
## dbl (4): prey_p_month, hrs_indoors, n_cats, age_years
## lgl (4): hunt, food_dry, food_wet, food_other
## dttm (2): deploy_on_date, deploy_off_date
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.cats.df<- cats_uk_referenceMake Histograms and Testing for Normality
hist(cats.df$prey_p_month, main = "Prey Per Month Histogram", xlab = "Prey (per month)")
qqnorm(cats_uk_reference$prey_p_month)
qqline(cats_uk_reference$prey_p_month)
hist(cats.df$age_years, main = "Age of Cat", xlab = "Age of Cat (years)")
qqnorm(cats_uk_reference$age_years)
qqline(cats_uk_reference$age_years)
Making histograms and qqplots of the prey and age data to evaluate the
normality of the data. The histogram of the prey data is very right
skewed and not normally distributed. The qqplot of this data is also not
normally distributed as it has this step-wise pattern and a majority of
the points do not align with the qqline. The histogram of the age data
is very right skewed as well and is not normally distributed. The qqplot
of this data is also not normally distributed and has a similar
step-wise pattern as the prey qqplot. Let’s see if a log transformation
makes this data a little more normal!
Transforming Data
cats.df$logprey <- log(cats.df$prey_p_month)
cats.df$logage <- log(cats.df$age_years)hist(cats.df$logprey)
hist(cats.df$logage)
Uh oh! The log transformation does not appear to fix our normality
issue. The data looks more normal than it did prior to the
transformation, but it’s still not great. Let’s try bining our variables
into categorical variables.
Because we are using our variables as categorical variables for the remainder of the analysis, we are unable to assess our data by the mean, medians, standard deviations, and standard errors.
df2 <- cats_uk_reference[-23,]
cats_uk_reference<- cats_uk_reference %>%
mutate(prey_p_month = case_when(
prey_p_month <= 3 ~ "low success",
prey_p_month >= 4 & prey_p_month < 10 ~ "moderate success",
prey_p_month >= 10 ~ "high success"
))
cats_uk_reference <- cats_uk_reference %>%
mutate(age_years = case_when(
age_years <= 3 ~ "Young",
age_years >= 4 & age_years < 10 ~ "Mature",
age_years >= 10 ~ "Senior"
))
levels(cats_uk_reference$prey_p_month)## NULLlevels(cats_uk_reference$age_years)## NULLNow, prey and age are grouped into 3 levels! These levels are in alphabetical order, let’s reorder them to be from youngest to oldest and lowest to highest hunting success.
is.factor(cats_uk_reference$age_years)## [1] FALSEcats_uk_reference$age_years <- as.factor(cats_uk_reference$age_years)
levels(cats_uk_reference$age_years)## [1] "Mature" "Senior" "Young"cats_uk_reference$age_years <- factor(cats_uk_reference$age_years, levels = c("Young", "Mature", "Senior"))
levels(cats_uk_reference$age_years)## [1] "Young" "Mature" "Senior"is.factor(cats_uk_reference$prey_p_month)## [1] FALSEcats_uk_reference$prey_p_month <- as.factor(cats_uk_reference$prey_p_month)
levels(cats_uk_reference$prey_p_month)## [1] "high success" "low success" "moderate success"cats_uk_reference$prey_p_month <- factor(cats_uk_reference$prey_p_month, levels = c("low success", "moderate success", "high success"))
levels(cats_uk_reference$prey_p_month)## [1] "low success" "moderate success" "high success"The variables are not recognized as a factor, so we will override these to be factors. Once they are factors, we can override the order using factor + the levels command. We recheck the levels to make sure they are reordered… Looks like it worked! Let’s make these variables into a table for further analysis.
catTable <- table(cats_uk_reference$prey_p_month, cats_uk_reference$age_years)
catTable##
## Young Mature Senior
## low success 26 43 12
## moderate success 3 5 0
## high success 7 3 1Table looks good and the variables are in the correct order! Now, we will run a Chi-Squared test of independence.
Statistical Test (Chi-Squared Test of Independence)
expected_proportions <- c(0.33, 0.33, 0.33)
sum(catTable)## [1] 100newexpected <- 100 * expected_proportions
chisq.test(catTable, p = newexpected)## Warning in chisq.test(catTable, p = newexpected): Chi-squared approximation may
## be incorrect##
## Pearson's Chi-squared test
##
## data: catTable
## X-squared = 5.5466, df = 4, p-value = 0.2357We chosen the Chi-squared test of independence because it tests between two categorical variables in order to determine if there is a significant relationship. We are testing this data against our null hypothesis: there will be no significant relationship between age and hunting success. Our alternative hypothesis is there will be no significant relationship between age and hunting success.
From this test, we get a X-squared variable of 5.5466 and a p-value of 0.2357.
Plotting a Mosaic Plot
mosaicplot(catTable, color = c("darkred","orange","yellow"), xlab = "Hunting Success", ylab = "Age", main = "Mosaic Plot of Prey per Month vs Cat Age")
Conclusion
Based on the chi-square goodness-of-fit test, there is no statistically significant evidence to suggest that the amount of time domestic cats in the United Kingdom spend indoors affects the number of prey animals they catch (X² = 5.55, df = 4, p = 0.2357). Since the p-value is greater than the significance level (alpha) of 0.05, we fail to reject the null hypothesis that age does not influence hunting success of domestic cats. This suggests that the observed distribution of prey capture does not significantly differ from what would be expected if indoor time had no effect.
This analysis is important because it helps to clarify assumptions about domestic cats’ impact on wildlife based on their habits. Understanding these patterns is valuable for ecologists and pet owners who are concerned about conservation efforts and responsible pet ownership.
However, there are limitations to this analysis. The sample size (n = 100) may not have been large enough to detect smaller effects. Factors like the cats’ hunting experience, age, or local prey availability were not controlled. This could influence prey capture independently of indoor time. Future studies could benefit from a larger, more detailed dataset or increased sample size, and consider other influencing factors, such as type of community hunted (rural vs urban vs suburban
References
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
McDonald JL, Cole H (2020) Data from: The small home ranges and large local ecological impacts of pet cats [United Kingdom]. Movebank Data Repository. doi:10.5441/001/1.pf315732
Pets Theater. (2022, February 15). Cats Catching Birds Video Compilation | Pets Theater. YouTube.https://www.youtube.com/watch?v=A93A2eORFnk