"increasing chances" and "necessarily leading to" are 2 different things. Consider for example smoking and cancer - among people who smoke the same amount of tobacco some will get cancer, some will not. That doesn't falsify the fact that smoking increases the chances of the cancer.
You link is actually enough here. From your link :
"Results: Between 12 and 17 mothers from the different groups accepted to participate, both groups of mothers of Somali origin had significantly lower values of 25-hydroxyvitamin D compared with Swedish mothers. "
So, i'm not sure what of the 2 well established facts you are arguing against :
1. Somalis in Sweden have very low D
2. Somali children born in Sweden have much higher autism rate.
Of course correlation isn't necessarily causation, yet the correlation (both - between populations as well as inside the Somali mothers population) is clearly stated in your link (have you read it yourself?) :
"Conclusion: Our findings of low vitamin D levels in Somali women entail considerable consequences in a public health perspective. The observed tendency, i.e. the lowest values in mothers of Somali origin with a child with autism was in the predicted direction, supporting the need for further research of vitamin D levels in larger samples of Somali mothers of children with and without autism."