Letter to the Journal of Applied Ecology

Executive Editor and Editorial Board
Journal of Applied Ecology
British Ecological Society
Charles Darwin House
2 Roger Street
United Kingdom

Dear Professor Milner-Gulland,

Regarding the recent Priority Contribution paper published in J. Applied Ecology
‘Foxes are now widespread in Tasmania: DNA detection defines the distribution of this rare but invasive carnivore”

Stephen D. Sarre, Anna J. MacDonald, Candida Barclay, Glen R. Saunders and David S. L. Ramsey

Please note that I am outraged that this paper was published as a peer reviewed objective scientific research paper (TT: Fox Invasion Threatens Wave of Extinction in Tasmania, here) when it has so many flaws and even the title is an unsubstantiated assertion. I have therefore written the following peer review for your consideration: -

1.  If a scientific research paper in a respected journal is to be peer reviewed, then the reviewer needs the raw data which has been manipulated in the complicated statistical analysis.  Yet the base data in the Sarre et. al. paper is totally missing apart from a few dots on tiny maps.  To make a proper review it would be necessary to know all the GPS coordinates, collection times, and associated fox DNA positive and negative scat information including the intimate details about how they were collected.  The peer reviewer from J Applied Ecology for this paper must be a genius to have positively reviewed this paper without any base data?  It is also a bit much to expect readers to accept their conclusions without data availability.

2.  If you are pregnant then you are unambiguously pregnant.  If you are a fox then you should be 100% a fox.  These things used to be absolutes.  Sarre et al. (2012) used scat mtDNA to determine that 56 were from foxes, and nearly as many (47) had fox like sequences but could not be definitely categorised as foxes.  It is not satisfactory, in my view, to introduce an arbitrary cut off line to decide just when a scat DNA analysis becomes a definite fox.  At the very least it is not appropriate to move the line, especially in the same supposedly scientific research paper.  Yet this is what they do because the 56 positive scats + 4 fox carcasses + a fox scull adds to 65 “confirmed fox presence” events for the statistical modelling study.  This compares with 61 fox positive scats reported by FEP previously.  The Bruny Island positive scat is included in the “confirmed fox presence” category (fig 4, page 8) but the map cunningly has the outline of Tasmania so faint that only a local would know that this is actually an island with 2 km of water between it and mainland Tasmania.  Pity they didn’t explain the discrepancies, and especially the difference between a definite fox mtDNA scat, a really definite one and a probable one? 

3.  The paper goes to lengths to supposedly prove by obscure modelling that foxes are widespread in Tasmania.  If the scat analysis was in fact correct, then this conclusion is obvious.  This is because:

        a).  for 15 scats to test positive as individual foxes, yet none of the 56 (or should that be 61?) scats testing as other than individual foxes, then it is mathematically nearly impossible for this result without about 10 times the number of fox scats found to be actually present but not picked up (only about 1 of 500+ likely fox scats was detected in any 12 week period during which scat DNA remains viable)

        b.)  for 56 fox mtDNA scats detected in 7658 general carnivore scats then 1 in 137 collected scats is a fox.  Extrapolating to the estimated 300,000 carnivores in Tasmania (a bit uncertain and possibly no foxes in some localities) but about 75,000 quolls, 75,000 devils and 150,000 cats - all broadly liking fox “core” habitat, and probably averaging approximately the same scat output as foxes (5 - 8 per day) - then that would relate to over 2000 foxes.

        c.)  The search for fox scats only tested 1 in 3 units in the various 3km x 3km cells and hence well over 300 fox positive scats would presumably have been found for an equivalent search of all the cells, plus a lot more if cells with less than 6% favoured fox habitat had been searched, and a lot more again if the search had been more thorough.  i.e. the 63 scats found would have been expected to be at least 1000 if the effort to find them had been expanded significantly.

Since most of the scats were collected between 2006 and 2010, the above results average about 4 years old, resulting in an expected number of foxes at the completion of 2012 of probably 10,000 assuming they doubled in number each year because of breeding. The poisoning effort, currently taking approximately 5 years to cover the “core fox habitat” just once would not have made much difference!

4.  Thanks to persistent and diligent detective work by the likes of Ian Rist, David Obendorf and Dr Clive Marks, the so called fox carcasses have all been revealed as very suspect, most likely hoaxing, and the single fox scull, conveniently found on the top of a stump at the back of Lake Sorell, has little credence as arising from a local fox.  The 2000 plus fox sightings are also not convincing unless sightings for thylacines, UFO’s, pumas in Victoria, Big Foot in the US and the Loch Ness monster are equally credible.  The overwhelming evidence for current fox presence in Tasmania originates from the fox mtDNA found in 61 carnivore scats out of over 10,000 scats collected, nearly all between 2006 and 2010.  If the scat evidence is correct, then Sarre et. al. are at least correct in asserting that they have probably reached the state of being impossible to eradicate.  If the DNA evidence is not credible, then the Fox Eradication Program (FEP) case for widespread occurrence of foxes in Tasmania evaporates.  Lack of fox photographs, despite innumerable camera traps, lack of hard evidence following investigation of reported sightings, lack of detectable lamb kill by foxes, lack of detectable fox sounds and lack of evidence of poisoned foxes, despite tens of thousands of buried poison baits, do not reconcile with the presence of probably 1000+ foxes, as suggested by an extrapolation of the scat DNA diagnostics.

5.  Since the bulk of the mtDNA fox scat evidence as detailed in the Sarre paper has been known for years, it is very coincidental that it has only just been published, just as the next round of FEP funding is due.  If the urgency for expanding the poisoning programme is as the paper suggests, then the delay in publishing and implementation of additional fox eradication efforts for some years was not helpful to eradication prospects.

The present poisoning programme has now covered most of the designated “core fox habitat” once, with some local “hot spots” poisoned more than once.  Since any fox poisoning in Tasmania using buried Foxoff 1080 baits is unlikely to poison more than about 50 - 60% of foxes that are in the poisoned area in any single baiting program, and since the systematic and extensive poisoning has now been deployed for several years, then any continuing detection of fox DNA scats is surely an indication of ineffectiveness, not a mandate to significantly increase the poisoning effort?  The data presented by Sarre et. al. Indicates that poisoning is not working and that a new approach is needed.  Unfortunately the FEP seems singularly set in its ways, rebuffing or ignoring any suggestions from outside its ranks.  For example I painstakingly crafted a 9 page draft report “  2012 Review of the Tasmanian Fox Eradication Program “ including lots of ideas for improving fox detection, which I submitted to FEP for comment in January 2012.  Despite repeated promises from the FEP General Manager that he would obtain a response from the FEP Technical Advisory Panel, he had to eventually advise me that they were too busy to respond.  Other independent scientists have repeatedly requested detailed information pertaining to scat collection details such as locations, times and collection details.  Such requests are invariably denied until FOI laws compel them to release limited information.  Provision of useful FEP information is becoming scarcer by the month and the dedicated FEP newsletter “Eradicate” has barely mentioned foxes in the last 2 issues.  Even the never changing map of fox presence dominating each Eradicate issue doesn’t present the scat data in a detailed manner – just 26 x 3 vague leaf shaped icons supposed to represent 61 fox scat locations, each covering hundreds of square kilometres. 

6.  With the fox evidence from scats revealing 61 scats from over 10000 collected carnivore scats (FEP data) = 0.6% or 56 fox scats from 7658 scats (Sarre et. al. data) = 0.73%, then just a small error component in analytical mtDNA analysis could totally account for the fox positive results.  The almost certain false positive result from the Bruny Island DNA positive scat, the use of minimally trained volunteers for much of the scat collection and the possibility for sample contamination from over 1000 fox scats introduced to Tasmania from NSW by FEP suggest some error potential.  It would be embarrassing and disastrous to extend the poisoning programme with the attendant costs and non target iconic wildlife collateral damage poisoning if all the fox scat data was an error or a hoax. 

One fortunate consequence of the determination that no individual fox was responsible for more than one mtDNA scat is that it indicates that sample contamination is not common.  For example if a collector spread material from a positive fox scat to another non fox scat then it would presumably have the same genotype and hence be from the same fox. Unfortunately only 15 of the 61 fox positive scats have been able to be identified to the individual fox level, and the total lack of more than one fox positive scat coming from the same fox is itself very problematic.  Unfortunately also the rigor the Canberra laboratory analysing samples applies to its analytical procedures to prevent and recognise contamination does not extent to the processes of sample collection and provision of procedures to automatically recognise both false negative and false positive results. False negative results could have been routinely quantified by regular inclusion of a known fox scat every few hundred general carnivore scats submitted to Canberra.  False positives are less easily detected, but the keeping of duplicate samples of all scats by FEP and the submission of a second sample, suitably disguised as to identity, presented for analysis after the first sample tested positive might have enhanced confidence in the positive result.  Duplicate analysis of samples, especially tentative fox mtDNA positive samples, by another laboratory, completely independent of the FEP, would also have been prudent.

The Achilles heel of fox scat use though is that it is so easy to hoax results.  Any traveller waiting to catch the ferry to Tasmania would only need to fill in waiting time by wandering the Melbourne docks to pick up scats which are readily carried in a pocket and distributed in the Tasmanian bush.  An anonymous phone call to the FEP that a definite fox had just been sighted in the area would ensure an investigation.  I think Sarre et al should appreciate that the world is not like a university ivory tower with well educated, polite and honest citizens.  Even the universities, law firms, political parties and bankers have their share of con men, fraudsters, thieves and generally unscrupulous employees.  There are people in the multitudes who would not hesitate for a minute to introduce fox scats to Tasmania if they thought it would make a good prank, assist a friend, relative or pub companion to keep a job, or just as payback to the government for a perceived grievance.  People with a vested interest in maintaining the fox poisoning program include not just the direct employees of the FEP, supply providers and analytical services, who are all undoubtedly above board in their actions, but also their associates.  It would only require a single person with the most frivolous motive to have provided all the fox positive samples found so far.  A slightly more challenging task (but not much) would be to dig up a fox bait (available at your local Hobart suburb lately), take it to the mainland along with some rufous wallaby meat, (only found in Tasmania) and poison your own fox.  (Or plenty of fox road kill there but nice to have poison in its system when it is found!) Then easy to bring it back to Tasmania (car, fishing boat or yacht) and plant it behind the latest poison front where it can be hailed as proof positive that foxes are in Tasmania and that the poisoning is working. 

7.  Regarding the fox habitat suitability modelling section in the paper.  Sorry but I think this section should be deleted.  Superficially it looks impressive, but arcane statistical twisting of suspect raw data is not going to provide useful information.  Correlating possible fox presence with vegetation type is especially dubious because broad vegetation categories in Tasmania are highly variable with fox favourable pockets in all general vegetation types, even rainforest and button grass.  The literature suggests that foxes are highly adaptable and are unlikely to be limited to general vegetation types and especially not to land designated core fox habitat by FEP.

Yours faithfully,

Ivo Edwards

*Ivo Edwards is a research scientist with a current interest in the development of improved, economical and humane soft fabric traps for Tasmanian Bennett’s and rufous wallabies, possums and feral cats.