Peter Apps1; John Moncur2; Scott J Campbell2
1Botswana Predator Conservation Trust, Maun, Botswana; 2SpectralWorks Ltd, The Heath, Runcorn, Cheshire, UK
First Published: BMSS, AZ Alderley Park, 2014
African Wild Dogs (Lycaon pictus, Figure 1.) are classified as Endangered by the IUCN, Their numbers are dwindling as habitat is taken over for settlement and agriculture. They live in packs with home ranges of several hundred to a few thousand square kilometres. When they range outside the borders of protected areas into livestock areas they come into fatal conflict with livestock owners.
The overall project aims to replicate the chemical signals that African wild dogs use to advertise their occupation of an area, and use the artificial scent marks to deter free-ranging wild dog packs from moving into livestock areas.
Methanol extracts of 65 urine samples from dominant African wild dogs were analysed by GC-MS. Qualitative and Matrix data analysis procedures were carried out using chromatographic deconvolution software. With optimum data processing settings the software does not generate false positives and the rate of false negatives is low and can be controlled by setting the signal:noise parameter. If components with closely similar spectra elute within the same retention window, the MatrixAnalyzer software assigns multiple target components to single peaks, which is preferable to components being missed.
Botswana is one of a handful of countries whose protected wildlife areas are large enough to sustain wild dog populations, but even here the dogs’ propensity to roam over huge distances takes them out of protected areas and into livestock areas where they come into fatal conflict with humans. Predators in livestock areas threaten peoples’ livelihoods, and the major cause of wild dog deaths is humans, so keeping the dogs and humans apart will benefit both sides in the conflict.
Bioboundaries will tap into the chemical signals that wild dog packs use to inform other dogs that an area is occupied and to stop packs from trespassing into one another’s home ranges. Artificial territorial boundaries that mimic the effects of natural pack territory markers will stop wild dogs inside protected areas from trespassing into livestock areas.
The challenges posed by de-formulating African wild dog scent messages are very similar to the challenges posed in the rapidly developing field of metabolomics – both need to find small differences in composition between very complex chemical mixtures, and identify the compounds responsible.
Figure 1. African Wild Dog African wild dog
Materials and Methods
Urine samples were collected from the BPCT’s African wild dog habituated study packs.
Urine-soaked soil was air dried at room temperature (approximately 22 °C), then sieved through a 2mm mesh. Approximately 3 g was accurately weighed into a pasteur pipette plugged with a small pledget of deactivated glass wool. Methanol (Sigma Aldrich Chromasolve) was percolated through the sample and the first 1 ml collected.
|Instrument:||Bruker 460GC / Varian (Bruker) 320MS|
|Column:||30 m x 0.32 mm x 0.5 micron RtxWax (Restek)|
|Carrier Gas:||Helium. Constant linear flow 35 cm/s|
|Injection:||Split-less. 2 µl. 250 °C|
|Oven Program:||2 minutes 50 °C, 5 °C /min to 240 °C for 10 minutes|
|Scan Range:||m/z 33 – 350|
Data analysis and peak deconvolution was performed using AnalyzerPro®
Results and Discussion
A typical GC-MS TIC is shown in Figure 2. Overlapped peaks cannot be properly quantified or identified, and most failures to find an MS library match are due to mixed spectra from co-elutions and only a few are compounds which are not in the typical libraries used for GC-MS. Deconvolution with AnalyzerPro unravels the complexity of the minor peaks of the dog scent chromatograms. We have been able to create a set of comparable data from these 65 samples for some 862 components.
Comparisons of results from wild dogs with those from other mammals have shown that none of the compounds identified so far in wild dog urine are unique to wild dogs, and the major peaks on a total ion chromatogram are common metabolites that occur in the urine and faeces of other mammals, so they are not likely to be signalling compounds. All the interesting things are happening among the minor peaks, produced by pico- to nanomolar gas phase concentrations which are easily within the range of a wild dogs’ sense of smell but close to the limits of detection for GC-MS analysis.
Mammal odours are chemically very complex and operate at very low concentrations, so chromatograms are crowded and peaks range in size over 4 – 6 orders of magnitude, with the smaller ones at the limit of detection of the analysis.
Figure 2. TIC of Methanol Extract of Male African Wild Dog Urine
Working out which of the hundreds of compounds are signals requires chemical composition to be correlated with signalling function – at the basic level we are looking for chemical features that are common to materials that send a particular signal, and absent from materials that do not send that signal. Such chemical features can be anything from the presence or absence of a single compound to some particular ratio between the concentrations of a large number of compounds. In either case we have to have a way of comparing the abundance of compounds between samples. Since the complexity is beyond the separating power of gas chromatography we have to be able to measure the areas of peaks that are overlapped to varying degrees in the chromatogram. Once we have an area for a peak it has to be compared to the areas for the same peak form other samples. The first part is deconvolution which we have established so far. The second part is the target listing, and list searching, which is much more of an optimisation challenge and will require further analysis.
To progress with identifying the African wild dog signals we require biological information about how the dogs respond to natural marks. The field research is particularly taxing and it may be possible to make some inferences from the study of domestic dogs.
We have already converted 65 chromatograms of wild dog urine volatiles into a set of comparable data on 862 components that are now able to be subject to multivariate statistics. Subsequent work will include running pairwise comparisons of wild dogs with leopards, lions, spotted hyaenas and cheetahs. The components that correlate with signalling will be identified and formulated into prototype artificial scents that the BPCT field team will test on their study packs.