ECOS 38 (2)

Monitoring populations of endangered species has been revolutionised by the availability of conservation technology but how does it affect life for those at the coal-face of data collection and analysis? 

The Kenya Rangelands Wild Dog and Cheetah Project (KRWDCP) studies the wild dog population of Laikipia County in Kenya. The project was created by Rosie Woodroffe back in 2001. Although much of the data collection involves many hours of on-the-ground fieldwork and observation, conservation technology has been central in contributing to over 15 years of this science.

New and improved data collection

Ecological research is increasingly moving away from traditional observational methods of data collection towards making greater use of technology. This has numerous advantages compared to previous methods, although there are some drawbacks. There are various pieces of technology that we use during our fieldwork, but arguably the most important are GPS (Global Positioning System) collars. KRWDCP first used GPS collars to study the Laikipia wild dog population in 2003, and means the project now has a high-resolution, long-term, dataset of wild dog behaviour data.

The GPS collars we use have three main components: a GPS unit, an accelerometer and a VHF (Very High Frequency) radio-tracking. Each of our study packs has one member fitted with a GPS collar which means we are able to collect data on the daily lives of our study animals in much higher resolution than we could ever achieve without them. The collars’ GPS units record their location at pre-programmed times throughout the day. These data mean we can see exactly where our study animals have been to within a few metres. This fine scale data allows us to pinpoint key events in the wild dogs’ lives such as the date that a litter of pups is born, where their den is, and then, a few months later, exactly when the pack leaves the den. The accelerometer is similar to those found in a smartphone or fitbit, and it allows us to see how active the dogs have been throughout the day – something that is almost impossible to accurately measure using traditional data collection techniques. We use VHF beacon along with an antenna and receiver to find our collared animals on the ground. In this way, we are able to find the dogs more regularly, allowing us to track the internal pack dynamics and follow each individual dog from as soon as it leaves the den right into old-age. This has been pivotal in allowing recent research into wild dog demographics.1

As well as increasing the amount of data collected, the collars also remove potential biases.  We can monitor the dogs in areas inaccessible to vehicles, as well as their activity levels and behaviour during the night. One example of the incredible data is given in the two maps below which show data from a young wild dog who left her natal pack shortly after being collared. One map shows reported sightings of the dog to the project, the other shows the data downloaded from her GPS collar. This illustrates how much more detail is provided by these collars when compared with following individuals on the ground.

Maps showing data collected about one wild dog's movement patterns from on-the-ground sightings (panel a) and from its GPS collar (panel b) across the same five month period. The green dots in panel a are locations of opportunistic sightings, red dots are locations of sightings where the VHF tracking function of the collars was used to find the dog.

Recently another key advantage of monitoring using collars has emerged - they enable us to keep monitoring the dogs even when it is not possible to be there on the ground every day.  In recent months Laikipia has been experiencing land invasions, looting and property destruction by armed herdsmen and their livestock, with the resultant clashes causing the deaths of several people in the area. This has meant that there are parts of our study area where it is not currently safe to go, but we are nonetheless able to continue to monitor our study animals without putting our staff at risk.

GPS collars – pros and cons

The use of collars to monitor individuals and populations, however, has become increasingly controversial. Recent articles2 have discussed the trade-off between the benefits of researchers using tags and the risks of the general public exploiting the tags to track and potentially harass animals. It is a difficult balance to strike. Ecotourism is an important source of income for Laikipia and many of the conservancies, where the majority of the area’s wildlife lives, rely on tourists to keep viable. Wild dogs are an important draw for many tourists. At some camps the majority of their guests visit mainly because they have a high chance of seeing wild dogs. Part of the reason for this is that our project works closely with our local collaborators, sharing information about the dogs in exchange for photographs and information about their sightings. These tourist sightings help us to track individual animals, and once alerted us to a dog caught in a snare, meaning we were able to find a vet to remove the snare. For a species that, within living memory, was considered vermin, and actively persecuted across most of Africa as a result, this positive change in public opinion is vital to the survival and fortunes of wild dogs , and has no doubt been aided by the fact they have become a valuable ecotourism asset.

Using technology in the field

When choosing which collars to use our highest concern is always to find models which will have the least impact on the animal, as such we’re always very aware of both the overall design of the collar and the weight. Best practice recommendations are to keep the collar weight to less than 3% of the animal’s body weight, although we aim for closer to 1% for our study animals. As the wild dogs in our study area average about 22kg this doesn’t leave us much weight to play with. Luckily for us the evolution of technology happens very quickly and so whilst the newest collar tech is usually not available to us, it rapidly becomes miniaturised to the point that we are able to use it. In older GPS collars you had to retrieve the collar to download your data, meaning there was always a risk that if you lost the dog (or the collar malfunctioned) you would lose the entire dataset. Thankfully these days we are able to download the data using a VHF connection, although even this becomes tricky when packs are hidden away at a den.

Getting collars onto study animals has proved challenging at times. Replacing or removing collars can be planned in advance, as you are almost guaranteed to be able to find a collared animal, but getting new collars out is much more difficult. Our study area is primarily thick acacia bushland which can make finding uncollared dogs a bit like looking for a needle in a haystack. The wild dogs only need to be a few metres away from you for them to be almost invisible. If you do find an uncollared animal, securing a vet who has time to come out and dart them can be a challenge in itself. Our local Kenya Wildlife Service vet covers a large section of the country which contains a lot of important wildlife areas, meaning he is incredibly busy and often isn’t able to attend.

As wild dogs live in packs the project only needs to collar one dog to be able to monitor what the whole pack is doing. Photo: Helen O'Neill

New frontiers in data analysis

The large amount of data obtained from GPS collaring allows for analyses not possible with many other ecological datasets and allows other studies to emerge from the project including from its early data. Large, detailed, datasets like that generated by the KRWDCP allow for research into increasingly complex research questions, ones which could not be answered using smaller scale observational data. An example is the project’s recent work looking into the effects of climate change on wild dogs.

Advances in technology have made processing and analysing data easier than ever. Open source software such as R and QGIS have opened up tools to ecologists that previously they often did not have the funding to get hold of, as non open-source equivalents such as Matlab and ArcGIS price many research institutes out, especially within ecology where funding is so sparse. Whilst Excel may not be an option for very large datasets – these tools are helpful replacements for data manipulation, and can be used for just about every analysis an ecologist could need to carry out. As a result of this, quantitative analysis in ecology papers has improved. Many biologists are now taught to use R and GIS as undergraduates, meaning they can carry out increasingly sophisticated analyses by the time they reach the stage of being a postgraduate student or researcher. Without these resources researchers on the project would not be able to build and run the models that we use in our research every day.

Too much data?

Having a large amount of data is an advantage, but it presents some interesting challenges, and sometimes unforeseen consequences. A number of the datasets generated through the project have millions of lines of data, which can make traditional methods of data manipulation challenging. For example Excel will frequently crash and die when these datasets are opened, occasionally followed by the whole computer. On top of this, time and resources are required to manage these large datasets, something that requires a high level of organisation and increasingly large amounts of space for data storage.

It is also crucial that staff and students working on these datasets have the skills to analyse them properly. Recently in ecology we have seen a rapid shift towards more quantitative skills. Ecology PhD students, and in many cases Masters and undergraduate students,  are expected to do all of their own, increasingly complex, data analysis. Gone are the days where the majority of ecologists teamed up with statisticians to analyse their data. This is an advantage of technology as everyone on the project now has very good data management and quantitative skills that will benefit them in the future. But rapidly learning a whole new set of skills, on top of, in many cases, extended periods of fieldwork, is difficult in the restricted time-frame of a PhD.

Radio tracking an elusive wild carnivore. Photo: Sarah Durant

Funding the legwork, the technology, and the analysis

Each GPS collar costs thousands of pounds. Raising money for long-term monitoring projects is always daunting3 and buying new, or even just refurbishing old, collars always takes a large chunk of a project’s annual budget. There is a misconception that projects working on large charismatic animals such as wild dogs always have ready access to funding but this is not the case. Ecological monitoring projects are not cheap to run and continually securing short-term grants to keep long-term projects running is difficult, despite the awareness that they are extremely important to both science and conservation.3

There are also struggles to fund the data analysis strands of the research. All these complex analyses, and the need to store and manipulate large datasets, means that increasingly sophisticated computing equipment is needed. It is tough to raise the money for fieldwork, but the actual computers with enough processing power to build and run models using these large datasets can be a harder sell to funders, as they are a lot less glamorous.

Things may be changing however, as funding streams and PhDs on offer are often increasingly geared towards more large scale, quantitative questions, and we are increasingly taking students from a non-fieldwork background. A recent PLoS Biology article said that “all biology is computational biology” highlighting the movement towards making the subject more mathematically and statistically driven.4  This is not by any means a bad thing. However, a risk for the future of ecology is that the fieldwork is being increasingly undervalued and underfunded - after all, why take the risk of sending someone out into the field when there are now so many large datasets available online? The data collection done throughout the history of KRWDCP (as with any long term project) took a large amount of time, money, blood, sweat and tears to carry out. This can often be forgotten, and funding for continued data collection is increasingly overlooked for funding projects analysing existing data. This risks circular research where the same data is analysed repeatedly, whilst becoming increasingly out-of-date. Fieldwork skills are essential in ecology, but there is growing concern that they are being undervalued or overlooked in favour of data analysis skills.

Wild dogs are co-operative breeders. The alpha pair breeds and the rest of the pack help raise the young, with one member babysitting the pups while the rest of the pack hunt, and bring food back to the den. Photo: Helen O'Neill

New collars for new data   

The evolution of technology continues and is opening up lots of exciting new research opportunities. A new generation of small Iridium GPS collars are becoming available which have the potential to revolutionise the way we study some aspects of wild dog behaviour.  These collars upload their data via satellite each day meaning that it is possible to monitor what the dogs are doing remotely. These collars will also make day-to-day monitoring of packs much simpler as we will know the areas they are in each morning, cutting down search times.

Our project is currently teaming up with the Laboratory for Animal Movement lab at Swansea University to fit collars kitted out with tri-axial accelerometers. These collars measure activity constantly, on three axes, providing movement data at a high enough resolution that you can see every footstep the collared wild dog makes. A phenomenal amount of data is produce in the process, on a scale not seen before on our project, meaning a super computer is needed to visualise and analyse the data. Similar collars have already revealed new information on wild dog hunting behaviours. The hope is we can examine fine scale behaviours in our study animals and that this can further reveal wild dog responses to temperature and other environmental variables.

Now we just need to raise the funds to buy and fit these state of the art collars…

References

  1. Woodroffe, Groom and McNutt (2017) Hot dogs (in press)

  2. Cooke, S. J., Nguyan, V. M., Kessel, S. T., Hussey, N. E., Young, N. and Ford, A. T. (2017) Troubling issues at the frontier of animal tracking for conservation and management,. Conservation Biology.

  3. Hughes et al. (2017) Long term studies contribute disproportionately to ecology and policy. Bioscience.

  4. Markowetz, F. (2017) All biology is computational biology, 2017, PLOS Biology.

 

Helen O’Neill 

PhD student at ZSL’s Institute of Zoology and UCL’s Centre for Biodiversity and Environment Research. Helen studies cheetah and African wild dogs in human-dominated environments.

Contact the author

Dani Rabaiotti    

PhD student at ZSL’s Institute of Zoology and UCL’s Centre for Biodiversity and Environment Research. Dani studies the effect of climate change of African wild dogs.  

Contact the author

Leave a Reply