dog population census methods

[Cynthia Lauren Mills]

I am involved in the designing a census method for the island of Rarotonga of the Cook Islands.  

I was hoping I could run my thoughts by all of you to get some input and criticism and maybe some help.  I will summarize my methods:

Rarotonga is a mountainous, volcanic island with a surrounding coral reef forming lagoons all around the island.  Human population inhabits the coastal lowlands and the inner, mountainous parts of the island are not inhabited and only partly exploited for agriculture.

Schematically I divided the island using a central point and radiating lines out to more or less turn the island into a pie.  I chose to divide on a basis of about 10 degrees (there were 34 regions, so the angle was a bit larger) as with test surveys I could do a section that size within 2 hours.  

I used a tourist map to delineate sections by landmarks and determined and recorded those landmarks using GPS.  Because there were no available maps that showed all the roads of the inhabited parts of the island, a complete description of the boundaries of a section was impossible to create in the time available.  So I made a  rule  that if a road started within the periphery (outer road) of a section, it would be included in its entirety in that section.  It turned out some sections were very large and took much longer than 2 hours to cover.

Dogs in Rarotonga are fairly connected to people's homes.  Although they are rarely contained (no fences, no leashes) they do stay close to the houses.  (There is governmental control of the dogs, consisting of Joe the Dog Man who goes out and shoots dogs.  He responds to complaints or just patrols to see if there are packs running about.)  As a consequence dogs seem to be keeping a low profile and not wandering much.  Walking the neighborhoods in the middle of the day and making some noise (or encouraging other dogs to walk with me) got most of the dogs laying around the houses to come out for counting.  

In addition, I (and other volunteers) approached people and asked about how many dogs lived in an area, such as a cul de sac or a compound.  Then I used the number in comparison with the number of dogs seen as an index to how many dogs were missed, typically.  So when I walked up a road I would count the dogs I saw and then ask someone how many dogs lived on that street.  I recorded the numbers and calculated a fraction:   3 of 6--I saw three dogs but six lived on the street.  

I recently calculated the population: I used the proportion of the island surveyed (about 85% of the island has been counted) and used that fraction to extrapolate the estimate for the whole island.  I used the survey index (I calculated I missed 0.48 of the dogs--this fraction was collected from 48 data points and I created a confidence interval from these points).  

I used the WHO/WSPA protocol, but have obviously deviated from some of the recommendations.  Since I was the sole survey person for much of it, I considered but did not ultimately use a sighting/re-sighting method to estimate an index for missed animals.  I was not sure I could remember all the dogs or effectively mark or photograph them when I was walking so many different areas.  I used a purely informal survey method, but I believe the people were honest in their estimates.  (The police restrict the number of dogs people can own to two, so if I asked how many dogs someone owned they would always say "Only two."  Instead I asked how many dogs lived in an area and made sure they knew that I was associated with the foundation, the Esther Honey Foundation, that takes care of animals there.)  

Thank you for any help you can give me.  

Cheers,

Cynthia Mills

Cynthia Lauren Mills

cynth...@mac.com

veterinarian and science writer

Award winning author of The Theory of Evolution, What it is, Where it came from and Why it works, and articles for Discover, Conservation, Bark and The Sciences.

[DogPopula...@googlemail.com]

Dear Cynthia,

Thanks for your message outlining the population survey you have
carried out in Rarotonga – we really appreciate your time in sharing
this situation and your experiences and I hope we can offer some
useful insights.

Our first thoughts were that you have obviously collected an extremely
valuable data set, especially as you have covered 85% of the island.
Counts can be used either for estimating the total population of an
area or as indicators for evaluating population changes over time.
You could use the counts from your observations (without any
extrapolation) as an indicator; by repeating this count, either
covering the entire 85% or a known portion, again in the future and
comparing these counts would provide a good reflection for how the
total population has changed (any increase or decrease in counts would
be assumed to reflect the same percentage increase or decrease in the
total population). You would, of course, need to be careful when
repeating counts to keep any confounding variables to a minimum
(weather, time of day, seasonality, etc). If you have kept detailed
records of locations of your counts you may also be able to show how
any changes in the population are distributed. For example there may
be ‘hotspots’ of increases or decreases in counts that may indicate a
location where the dog population has been particularly responsive or
non-responsive to your intervention. This is particularly useful
considering one of the main reasons for monitoring is to provide
information we can then use to help improve the effectiveness of our
interventions.

In your description you explain how you have used information to
extrapolate your data to provide a population estimate – by talking to
local people to develop an estimate for how many dogs you miss during
your counts. We were interested to hear your thoughts on how accurate
you think this will have been. Our concern would be that people may
define the question “how many dogs live in this street/area?”
differently. For example, did they understand what you meant when you
said this ‘street’ or ‘area’? Would their perception of the
boundaries of this street be the same as yours? Would they define
‘live on this street’ the same as you and their neighbours? Is it
possible that a single dog would be described as living on more than
one street because it feeds (either from garbage or purposeful feeding
from more than one referral household) on different streets at
different times of the day?

An alternative to using reports from local people would be to use a
mark applied as part of an intervention. Are dogs marked as part of
any Esther Honey Foundation interventions? If so, there is a
description of a method you could use in the guidelines (page 14).
This basically uses information on the number of dogs you have marked
since your intervention started and the percentage of marked dogs that
you see on your counts to provide an estimate of the visibility of
dogs and hence an estimate of the total population size. The
important point for any survey based on sight-resight techniques is
that marked and unmarked dogs need to have the same chance of being
seen your survey. If dogs that are more visible (e.g. if they are
more friendly) are also more likely to become part of your
intervention and therefore marked, you will get an underestimate of
the total population size, and visa versa if marked dogs are less
likely to be seen than unmarked dogs (e.g. as would happen if most
dogs that are marked through your intervention that live within
private areas such as houses or the grounds of businesses).

In conclusion, it is difficult to deal with this problem of some dogs
not being visible and hence your count missing some dogs. One option
is to just use the count as an indicator and not worry too much about
missing some dogs, so long as you consistently run the same chance of
missing these dogs on your repeat counts.

By the way, Google Earth has some good images of Rarotonga which are
detailed enough to show most roads, these may be useful for splitting
the island into regions and tracking your progression through those
regions.

> cynthmil...@mac.com