The link provided links back to this topic. Here's the proper link: http://www.sciencema...study-concludes
Linked to this, is another discussion which is developing at the moment, and which centers around the question if the one-off auction sale in 2008 led to a marked increase in poaching.
In 2011 a paper was published by Underwood et al which found it didn't: http://journals.plos...al.pone.0024165
Then in June a working paper was published by Hsiang, based on the same data, which found compelling evidence that there was a relationship between the one-off auction and an increase in elephant poaching: http://www.nber.org/papers/w22314
The authors of the original paper wrote a rebuttal to that, centered around methods:http://www.fmunderwo...g-of-elephants/
Hsiang wrote a rebuttal to that: http://www.g-feed.co...o-elephant.html
Underwood wrote a reply to that again: http://www.fmunderwo...ekars-analysis/
Hsiang replied to that again: http://www.g-feed.co...hting.html#more
Ain't it fun being in the scientific world?!
Some people from the Worldbank chipped into the discussion too: http://blogs.worldba...ing-be-question (link seem to be broken). They present a figure with median raw ivory prices in China and Africa and write "“it is notable that neither price series shows a marked change in 2008, casting some doubt on the claim that the 2008 sale had had any impact on illegal ivory markets". However, in the figure, there is a sharp decrease in price between 2008 and 2010 in both China and Africa. Before 2008 there isn't really a price increase in Africa, but after 2010 there is a clear increase. Similarly, the price increased slowly in China before 2008, and more rapidly after 2010.
The whole discussions between Hsiang and Underwood centers around the methods they use. It's a bit technical but I won't go into it here. What they don't address is the quality and usefulness of the data. While proportional data can be good to address questions like this (is there an increase in illegally killed elephants), that's not always the case.
For example, in the data, in 2007 in Samburu 97 elephant carcasses were found, and 23 of those were classified as being killed illegally, or a rate of 0.23. In 2009, in the same area 326 carcasses were found and 86 of those were classified as illegal, or a rate of 0.26. According to this analysis there is not really any difference. But if you were the manager of Samburu, when the number of poached elephants triples, you're worried!
Another indication that this data isn't really fit to detect poaching trends is shown by the following. The Selous' elephant population declined from about 39,000 to about 13,000 between 2009 and 2013. Meaning about 5-6,000 elephant carcasses must have been present in the Selous in those years each year. Natural death rates in elephants, as they are a long-lived species, are really low , ~0.6% (see: HERE). This would mean that in a population of about 39,000 elephants there would be ~230 natural deaths, and in a population of 13,000 about 80. Annually between 100 and 224 carcasses were reported. If there were about 5,000 illegally killed elephants, and about 200 naturally died elephants you would expect a ratio of about 0.96 of poached elephant, yet this ratio varies between 0.48.0.73. This indicates that many illegally killed elephants are classified as not illegally killed. This often happens when an elephant gets shot, doesn't die immediately but hours/days/weeks later and falls on the side with the bullet wounds. But apart from that, you could wonder what 200 carcasses can tell you about 5,000+ carcasses. Arguably such a small fraction isn't really a representative sample.
Another clue that things are not quite right with the data is the following. If a population is stable, and the birth rate is stable, than death rate should be stable as well. Most elephant populations are in decline, and birth rates are not really fluctuating. Yet, the data shows an increasing amount of elephants which died naturally. This is highly unlikely in stable or decreasing populations, indicating an issue with the data. This is further reinforced by two, known to be increasing populations (Chobe and Kruger) having relative stable numbers of natural deaths (they do vary, with a spike in drought years, but don't have increasing numbers of natural deaths over the reporting period). How can known increasing populations have stable numbers of natural deaths, while many probably declining populations have increasing numbers of natural deaths?