See: Sarra Elkamel et al
"Ancient and recent Middle Eastern maternal genetic contribution to North Africa as viewed by mtDNA diversity in Tunisian Arab populations"

http://onlinelibrary.wiley.com/doi/10.1002/ajhb.23100/abstract?wol1URL=/doi/10.1002/ajhb.23100/abstract&regionCode=US-NY&identityKey=093ea0d1-779c-411b-a5bc-6d8a9643a5a2

The entire paper is available:

"Abstract

Objectives

Through previous mitochondrial DNA studies, the Middle Eastern maternal genetic contribution to Tunisian populations appears limited. In fact, most of the studied communities were cosmopolitan, or of Berber or Andalusian origin. To provide genetic evidence for the actual contribution of Middle Eastern mtDNA lineages to Tunisia, we focused on two Arab speaking populations from Kairouan and Wesletia known to belong to an Arab genealogical lineage.
Materials and Methods

A total of 114 samples were sequenced for the mtDNA HVS-I and HVS-II regions. Using these data, we evaluated the distribution of Middle Eastern haplogroups in the study populations, constructed interpolation maps, and established phylogenetic networks allowing estimation of the coalescence time for three specific Middle Eastern subclades (R0a, J1b, and T1).
Results

Both studied populations displayed North African genetic structure and Middle Eastern lineages with a frequency of 12% and 28.12% in Kairouan and Wesletia, respectively. TMRCA estimates for haplogroups T1a, R0a, and J1b in Tunisian Arabian samples were around 15 000 YBP, 9000 to 5000 YBP, and 960 to 600 YBP, respectively.
Conclusions

The Middle Eastern maternal genetic contribution to Tunisian populations, as to other North African populations, occurred mostly in deep prehistory. They were brought in different migration waves during the Upper Paleolithic, probably with the expansion of Iberomaurusian culture, and during Epipaleolithic and Early Neolithic periods, which are concomitant with the Capsian civilization. Middle Eastern lineages also came to Tunisia during the recent Islamic expansion of the 7th CE and the subsequent massive Bedouin migration during the 11th CE.

I find myself in a situation where I rather agree with the conclusion but I don't think the authors have completely proved it. I don't think studying Andalusian populations or only Berber ones, which prior papers did, gave us an accurate picture of Middle Eastern mtDna gene flow into North Africa, but studying only Arabic speaking groups isn't going to give an accurate percentage either, in my opinion.

To really get this right I think they'd have to do very broad scale sampling. Also, they need complete mitogenomes.