Gavin Baker on the Journal of Insignificant Inquiry reports a spike in the growth of DOAJ in 2008, based on data generated from the DOAJ new title search page.
The DOAJ numbers are the best estimate I am aware of of how many open access journals there are in the world, however it is important to remember that DOAJ is intended as a directory of journals, and is not a perfect measure of OA journals.
Thanks, Gavin, for a useful addition to the Dramatic Growth of Open Access Series.
Some factors to consider when looking at DOAJ as a measure of open access journals:
Discovery - there could be open access journals which DOAJ staff are not aware of.
Vetting - DOAJ journals go through a vetting process; they not only need to be fully open access and peer reviewed (or equivalent), DOAJ needs evidence. Because of the vetting process, there is always at least some delay between when a journal becomes OA and when it is included in DOAJ.
Useful delay - DOAJ does not include forthcoming journals, and does not automatically and immediately include new journals. Journals must be active to be included in DOAJ; sometimes, it makes sense to wait for a few issues to be released.
Human resources - the people at DOAJ are human beings. Minor fluctuations in growth can simply reflect someone's vacation time. If libraries want DOAJ to be as up to date as possible - take out a membership in DOAJ so that this important service has the support it needs to keep up.
Language - while the DOAJ staff seem remarkably comfortable in a multilingual environment, it must add some complexity that OA journals can be in any language, in any alphabet. I have no idea whether discovery and/or vetting is slower for some languages than others, but it does seem likely.
Weeding - DOAJ removes journals that no longer meet its criteria for inclusion (fully open access, peer reviewed or equivalent, and active).
Journal start dates in DOAJ reflect the first OA issue, not necessarily when a new journal started, or when an older one converted to OA.