Hi Everybody
We are developing a framework for monitoring free roaming dog populations in the cities of Tierra del Fuego. The objective is to monitor the mid and long term success of the policies to reduce free roaming dogs in the streets of the cities, like neuter and responsible pet ownership. Cities are expanding in surface and adding new neigborhoods over the course of the years.
For that reason, and to avoid restrictions of design-based surveys, we are considering the use of Density Surface Models, also to be able to explain changes in dog presence by covariates.
Then, we should be able to tell the towns to follow a series of tracks (streets) that need to be surveyed over the years. Adding covariates, we should be able to develop a DSM to assess abundance based on these covariates.
Then, here comest the two questions:
ONE: The issue that comes to my mind and I want to share with you hast to do with modelling all over the space of the city, when free roaming dogs are distributed only along the streets. The, the streets are like a mesh of something like 21m wide, separated by the about 800 blocks of around 90m side.
Potential solutions I was thinkg were:
1. Model the data as if they were collected and build a prediction grid where blocks can´t hold dogs (as if they are "0") meanwhile streets has a value of "1". Then I wonder how much the smoothing all over the surface will affect the predicted results.
2. Model the data, but adding "dummy" transects located over some of the blocks, just to tell the model that inside the blocks there are no dogs... However the modelling may be affected by the spatial spread of these dummy transects.....
3. Use something like the soap film smoother but I can´t figure out how we can handle such amount of blocks...
TWO:
About detectability of dogs. When travelling along the streets we need to record dogs that can walk freely at the sidewalks or at the streets. Streets has parked cars along the sidewalks, and many houses do not have fence, so dogs can sleep inside a parcel but leave the parcel boundarie as soon as they want to bark or whatever. This impose visibility issues that affect detectability.
To reduce that I was thinking in make dogs more visible:
1. Survey the streets going behind garbage collectors, as these guys atract dogs either by their behavior (running along the streets) or leaving garbage behind ocasionally).
2. Broadcasting dog barks from the car meanwhile we go travelling. I know that this affect the results of a survey, but if we want to see dogs in the streets I understand that we need to make them out.
Any idea or previous experience in that is welcome. Many thanks
are there people who deliberately put out food for them, like some people do for cats?
Hi
If you have some time to do the research (let's say a year), i would think of making 'dog spotting' a kind of community asset.
Code a mobile App to send GEO data and maybe a photo of the dog(s) to a server. Have a website to display these photos, so people can see, if 'their dogs' roam. Start with a PR campaign involving local politicians and newspapers. Maybe go to schools to 'recruit' spotters.
Working with public will have the people discussing the neutering scheme too.
Andreas
(sorry for the telegram style, but i retyped this the third time now and get annoyed of the shortcoming of research gate frontend. Hope you got it basically.)
Hi,
As an exercise as part of a population estimation course, students at a population estimation course I taught applied closed capture recapture sampling and modelling to estimate stray dog numbers in an urban area in south India. It worked pretty well back then, and now there are advances in spatial capture recapture which will allow you to define dog habitat and non-habitat (1/0), explore fine-scale spatial variation in density, account for movement of individuals within and even out of the sampled area, among many other benefits.
This would work best with photography using hand-held cameras across the area of interest rather than fixed location camera traps. However, this brings up the issue of how you would analyse the data: you could use the SCR models developed for area searches, but for which coding would have to be done in BUGS/R. Or you could use mid-points of your search segments (as long as you keep them short) as approximations to fixed location camera traps (or other passive detectors) and use software such as SPACECAP (see link below) or DENSITY to analyse your data easily.
All the best!
Article Program SPACECAP: Software for estimating animal density usi...
Please follow my article Population of free-ranging dogs in West Bengal. 2001, in Acta Theriologica
- Badges
- Science topic
- Similar topics
- Biological Science
- Zoology
- Ornithology