INTRODUCING FOREST "BAD" RESEARCH
Forest Research, along with the Forestry Commission, published its latest guess work over the population of wild boar in the Forest of Dean.
We know it is guess work because they say so themselves in their own conclusion!
But this has not stopped them promoting a scientifically indefensible argument about numbers to gullible councillors and the public alike.
In what appears to be science, Forest Research's latest attempt to support the Forestry Commission's urgent need to increase meat sales is one of sheer incompetence and blatant cronyism.
Here is the link to download the Forest Research document that alludes to a wild boar population of 819.
Now, we will ignore the amateurish style of the document and its footer margin claiming these survey results are for 2013. These numbers are a result of transects made by rangers in vehicles between 17th February and 11th March 2014. The length of transect totalled 167.4km about an area of 66.4km2 (almost the entire Forest area).
It is a series of line transects the rangers have used for many years, principally to calculate deer numbers but now expanded to include wild boar too.
The transects start in the evening as darkness falls and take place every other night and continue for up to 8 hours per night over the period. The transect of 167.4km is a cumulative total made over various parts of the Forest during the 3 weeks. Each transect is passed over once.
They use sensitive thermal cameras to detect the boar and deer and put the observations into a program called "Distance", using the methodology according to Buckland et al (Oxford University Press, 2001).
DISTANCE SAMPLING
This methodology is known as "Distance Sampling". It is a method requiring great skill by the observers, for not only MUST they record EVERY boar seen on a transect, they must also ACCURATELY calculate its distance and direction from the transect using laser technology or using marked distances on the ground.
It is a statistical method of counting that alleviates any need to count every single boar in the Forest, but a REPRESENTATIVE SAMPLE that can be scaled up to the size of the entire Forest.
Of course, driving along even 167km of track does nowhere near allow visible checks on the entire Forest even IF the surveys were done in the daytime. So only a tiny (<<1%) fraction of the Forest is surveyed.
But the thermal equipment employed does allow the ranger to see very clearly if an animal is detected by a bright glow in the viewfinder. Of course, the further away the boar is, the smaller is the glowing object becomes. This often proves impossible to determine the difference between species of animal (say deer and boar) at even moderate distance from the vehicle. Additionally, animals beyond just 10 metres usually means the animal is in tree cover and undergrowth adding to the difficulty.
Indeed, the FC in 2011 argued that wild boar often hide in the bracken out of the view of their thermal cameras. These were the words of Ian Harvey in 2011 after that year's census found only 16 boar (scaled up to equal several hundred of course!)
Distance Sampling is a technique that actually grants bias to more distant observations - the more boar seen at greater distances, the greater becomes the population result. Can we trust rangers to observe small glowing animals in a viewfinder correctly after travelling a tiring 8 hours through the night?
According to Buckland (2001): "Sloppiness in detecting objects near, and measuring their distance from, the line or point has been all too common [in previous research]..{and] proper design and field protocol have not received the attention deserved."
Buckland (2001) states categorically that animals must be detected with CERTAINTY. "Its importance cannot be overemphasized." (Chapter 2)
OH DEAR, THE BAD RESEARCH GETS WORSE
This difficulty is the LEAST of Forest Research's problems (which they purposefully keep from you)!
Even a quick read through Buckland et al (2001) soon highlights several major requirements for the sample size to be anywhere close to accurate (Chapter 7):
1. A pilot study is required to calculate starting assumptions such as encounter rates, distance bias, average boar group size, ratio of group to lone boar, and so on.
2. Transects must NOT be along roads or tracks were animals may prefer to travel at night.
3. The animals must NOT MOVE when observed.
4. The actual observed number MUST be greater than 60.
5. The distribution and density of boar must be equally distributed throughout the entire areas of sampling.
6. Locations of groups of boar must be plotted accurately using a mid-point of the group.
7. Every boar on a transect MUST be seen and counted (no hiding in the bracken this time please).
8. Distance to the animal must be calculated accurately.
In all 8 requirements, Forest Research have either broken these rules or could not have complied with them.
Here's just a few blatant errors....
No pilot study has been done nor published.
Counting an individual more than once (including over successive nights
if used) obliterates the accuracy of the result, increasing the final
total. Well, well. Forest Research seem ignorant of basic boar behaviour!
Using tracks or roads is junk science for boar counting. Some boar prefer roads and tracks to move at night as witnessed by verge diggings.
Some boar will readily move towards an observer for a closer look or sniff. Boar moving closer reduces the final population figure. Many boar prefer to run, especially from a noisy ranger's vehicles (they are smelly and have association with guns) and therefore, will increase the population figure - and quite significantly too!
The wild boar in the Forest of Dean are NOT distributed equally - and even the Forest research document shows you this by way of a plot and its words even says so!
Average group numbers on any night cannot be averaged as this changes throughout the seasons. Assuming too high a group size, or numbers of groups (put at 41 in the report) will massively exaggerate the final total. In some calculations within the distance-sampling method, individuals are allowed to be counted as a group of a predetermined size - but this doesn't work for boar - every boar MUST be counted.
Forest Research admit in their report to calculating distance to boar by assessing the size of the boar! OUTRAGEOUS!
Even the dim-witted Dougal McGuire and Father Ted have learnt about the difference between baby pigs near to, and big fat pigs far away.
CONCLUSION
According to Forest Research's own document, it concludes by saying that "...clustering of boar has the effect of reducing the precision of the estimated density in the forest as a whole."
Sadly the very next sentence compares this year's scientific abomination to their 2013 guesswork for some form of mutual corroboration.
"The 2013 estimate is within 95% confidence limits obtained from the 2014 survey, indicating that it is plausible (although still unlikely) there has been no change in numbers AT ALL."
This last sentence regarding 2013 is both pseudo-scientific and intentionally misleading. 2013 was not a pilot study nor was it calculated using distance-sampling.
COUNCILLORS BEWARE
The Forestry Commission are increasingly relying on the gullibility and scientific ignorance of local councillors to support an unnecessary cull. The FC no longer care what the public think, they want the councillors to decide.
The descent from science into the employment of "Localism" as a weapon against the boar is far too transparent a tactic for us to be conned. This is the government treating us like idiots folks. Infiltration into committees by government stooges to win debates is nothing short of "behavioural adjustment by propaganda" -something that 1930s Germany also sleepwalked into.
Given that many requirements for distance-sampling could NOT have been met, and hence easily leading to vastly over-estimated numbers, the last sentence of Forest research's propaganda document should have stated (if there is any scientific integrity within Forest Research)..
Wild Boar numbers in the Forest of Dean cannot be calculated accurately using the method Forest Research have chosen, and boar numbers may have increased, stayed the same, or decreased compared to our previous guess based primarily upon a need to generate meat sales and roadside diggings.
Buckland would not be impressed.
No comments:
Post a Comment