Work Description

Data Set Title: Free-Roaming Dog Image Dataset (Tulum, Mexico)
Authors: Michael A. Lyons, Rumaan Malhotra and Cody W. Thompson

Suggested citation: Lyons, M. A., Malhotra, R., Thompson, C. W. (2022) Free-Roaming Dog Image Dataset (Tulum, Mexico) [Data set], University of Michigan - Deep Blue Data. https://doi.org/10.7302/yncp-6w10

Research Overview

Free-roaming domestic dogs (Canis lupus familiaris) pose major conservation and public health risks worldwide. To better understand the threat of domestic dogs to wildlife and people and add to the growing literature on free-roaming dog ecology, a study was conducted to estimate the dog population in Tulum, Mexico. A modified mark-recapture technique and program MARK were used to obtain dog population estimates along six different transects dividing the city. Population estimates ranged from 19.75 dogs in one transect to 101.841 dogs in another, with 150 total dogs identified throughout the study and an estimated minimum population density of 48.57 dogs/km2. Fecal samples were also opportunistically collected for parasite identification through fecal flotation analysis using the McMaster technique. Out of 25 samples collected, 19 tested positive for gastrointestinal parasites with the most common species found being Ancylostoma caninum, followed by Toxocara canis, Dipylidium caninum, and Cystoisospora spp. Parasite loads ranged from 50 to 10,700 ova per gram of feces. The large population of free-roaming dogs and the prevalence of three zoonotic parasites highlight the importance of understanding free-roaming dog ecology and educating the public on the health risks free-roaming dogs pose.

Los perros callejeros (Canis lupus familiaris) representan un gran riesgo para la conservación de animales y la salud pública mundialmente. Para comprender mejor la amenaza que significan los perros domésticos para la fauna silvestre y los humanos y aportar a la creciente bibliografía sobre la ecología de los perros callejeros, se realizó una investigación para estimar la población de los perros en Tulum, México. Se utilizó una técnica modificada de marcado y recaptura junto con el programa MARK para estimar la población canina en seis transectos de la ciudad. Los estimados varían desde 19.75 perros en un transecto hasta 101,841 en otro, con un total de 150 perros identificados en el transcurso de la investigación y una densidad mínima estimada de 48,57 perros/km2. Además, se hizo una recolección oportunista de muestras de heces para la identificación de parásitos por medio del análisis de flotacíon fecal, con el método McMaster. De las 25 muestras recolectadas, 19 resultaron positivas para parásitos gastrointestinales, de las cuales las especies más comunes fueron Ancylostomoa caninum, seguida por Toxocara canis, Dipylidium caninum, y Cystoisospora spp. Las cargas parasitarias variaron desde 50 hasta 10.700 óvulos por gramo de heces. La alta población de perros callejeros y la prevalencia de tres enfermedades zoonóticas resaltan la importancia de entender la ecología de los perros callejeros y educar al público sobre los riesgos que significan los perros callejeros para la salud.

Methods

Modified mark-recapture methods.---To estimate the population of free-roaming dogs in Tulum, Mexico, a modified mark-capture-recapture technique was implemented. Instead of capturing and physically marking individuals, a non-invasive approach using digital photography was used to photograph dogs with a Samsung Galaxy S9 phone camera to identify individuals and match “recapture” events. Seven transects approximately 475 meters apart were plotted across the city of Tulum. Transects ranged in length from 370 meters to 2,949 meters. Everyday three transects were chosen randomly to be surveyed.
To “catch” dogs at different times of day, transects were biked around 5:30 PM on Mondays, Wednesdays, Fridays, and Sundays, and around 9 AM Tuesdays, Thursdays, and Saturdays until each transect was surveyed four times in the morning and four times in the evening. Since biking three transects per day took around 2 hours total, the order that the randomly selected transects were surveyed was modified daily so that individual transects were not always sampled at the beginning or end of the 2-hour sampling period. For each transect, photographs were taken of every dog encountered unaccompanied by an owner. When possible, multiple photos of single encounters were taken to aid in dog identification. Each photo was timestamped and included geographic coordinates. Sex was recorded when it was easily determined, and notes were taken on whether individuals were seen previously.
Individual dog identification.---Every dog encounter, defined as a picture being taken of a dog, was given a unique collection number. For example, in a sequence of two pictures, one with one dog and the other with two dogs, there would be one collection number corresponding to the first photo and two more unique collection numbers corresponding to the second photo. As each photo was processed, collection numbers were given to each dog encounter, and the date, time, latitude, longitude, transect number, sex, and dog color were recorded. Additionally, each dog was given a unique number as an identifier for the individual starting at number one. Dogs were identified using a combination of traits, such as sex, coat color and pattern, tail size and color, and collar type and color when present. To ensure the same individual was not identified as two different dogs, every dog in every photo was compared to all dogs with the same color in previously processed images, all previously encountered dogs on the same transect, and all previously encountered dogs on adjacent transects.

File Inventory

All images are labeled with the acquisition date and a unique identifier generated by the camera, e.g., 2021.0208_180616. Metadata for each image is located in the Excel file, FRD_Full_Data.csv. Available metadata include a date and time of acquisition, location information, and information on the dog being imaged.

Definition of Terms and Variables

Collecting_Event_No.---The sequential number for the collecting event.

Photo_No.---The image number from the camera used for dog identification. Collection events that did not have an image are included and indicated by “No Photo”.

Photo_Label.---The name assigned to the image used for dog identification.

Date.---The date of image collection.

Time.---The time of image collection.

Transect_No.---The transect that the imaged dog was collected.

AM/PM.---The time of day that the transect was sampled.

Latitude.---The latitude of the image in decimal degrees.

Longitude.---The longitude of the image in decimal degrees.

Total_Dogs.---The number of unique dogs in the image.

Dog_ID_No.---The unique number assigned to an individual dog.

Sex.---The sex of the dog imaged. Sex is identified as male, female, or unknown.
Color.---The color of the imaged dog.

Size.---The size of the imaged dog.

Additional_Characteristics.---Other characteristics of the imaged dog.

Comments.---Additional comments for the image.

User Access

Each file is a “jpg”, and is viewable with a standard image viewer on computers.