It's not necessarily that the study was retracted. The news article from the Guardian you linked mentioned that the study was submitted to the journal Personal Relationships; this means it had not yet been accepted for publication. And indeed it looks like that study never got published there despite all the media coverage.

Actually it has finally come out, 5 years later! Burton-Chellew, M.N and Dunbar, Robin I. M. (2015). Romance and reproduction are socially costly. Evolutionary Behavioral Sciences, 9(4), 229-241. http://dx.doi.org/10.1037/ebs0000046

From the abstract

We used an Internet sample of 540 respondents to test and show that the average size of support networks is reduced for individuals in a romantic relationship. We also found approximately 9% of our sample reported having an “extra” romantic partner they could call on for help, however these respondents did not have an even smaller network than those in just 1 relationship. The support network is also further reduced for those who have offspring, however these effects are contingent on age, primarily affecting those under the age of 36 years. Taking into account the acquisition of a new member to the network when entering a relationship, the cost of romance is the loss of nearly 2 members. On average, these social costs are spread equally among related and nonrelated members of the network.