While "South North Africa" (the Sahara belt) is the desert it was for millennia, "North North Africa" (Atlas Mountains + coastal land) is still quite capable of agriculture. Hundreds of thousands French farmers moved to areas like the Mitidja plain around Algiers in the colonial era, and while independent Algeria has been a net food importer I think this is more due to recent changes (rural exoduses to the cities, increasing urban population requiring more food, civil war damage, etc.) rather than lacking the basic potential for productive agriculture. I can't speak for Morocco but I assume the situation is similar. I think Tunisia is still a heavyweight in olive oil exports, Italian olive oil is sometimes just Tunisian oil bottled in Italy.
Algeria's productive lands/climates are mostly concentrated on a narrow strip of land close to the coast or in the cool mountain highlands, but is that really such a divergence from the pre-Muslim era? The
old centers of the Numidian kingdom were all quite close to the coast, and I assume Numidia's the breadbasket people talk about when discussing Roman North Africa. St. Augustine's homeland of Tagaste and workplace of Hippo Regius were both in coastal Annaba province. The area that was fertile then is, for the most part, the same area that is fertile (if punching under its weight) now.
That area has frayed at the edges since 600 AD, and the introduction of new pastoralists + decay of Punic/Roman irrigation systems likely worked alongside climactic shifts to cause that. However, outside of aberrations like the Banu Hilal invasion I don't think the Arab rulers of the area were really that hostile to or ignorant of sedentary lifestyles. They probably just know enough about the types of systems used there. A similar thing happened in Persia for a bit during the Turkish and especially the Mongol invasions, but as updated lists of canals and manuals on water maintenance became more common so did recovery/reconstruction efforts.
Sorry for the lack of sources throughout, but the point I'm trying to make is that when we talk about "what North Africa is today" we're talking about a bunch of disparate regions that barely seem like they belong in the same continent, much less a neighboring triplet of countries. It's important that we don't end up comparing Roman North North Africa to modern South North Africa.