Yes that blue line is retractions in scientific journals. Dr Roger Pielke Jr. notes on his blog that:
The Wall Street Journal reports that retractions of scientific papers have surged in recent years, with the top 3 journals issuing retractions being PNAS, Science and Nature. The graph above shows the increase in the rate of retracted papers.
Pharmalot provides a summary:
[T]here were just 22 retraction notices that appeared in journals 10 years ago, but 139 were published in 2006 and by last year, the number reached 339. Through July of this year, there were a total 210 retractions, according to Thomson Reuters Web of Science, which maintains an index of 11,600 peer-reviewed journals.
Meanwhile, retractions related to fraud rose more than sevenfold between 2004 and 2009, exceeding a twofold rise traced to mistakes, according to an analysis published in the Journal of Medical Ethics. After studying 742 papers that were withdrawn from 2000 to 2010, the analysis found that 73.5 percent were retracted simply for error, but 26.6 percent were retracted for fraud. Ominously, 31.8 percent of retracted papers were not noted as retracted (read the abstract).
It should also be noted that there are more journals now than in 1977 and we have rapid publishing tools instead of a drafting table and a typewriter.
I wonder if the Monnett polar bear paper will become a new data point on the graph above? It certainly started out as one.