Arbuscular mycorrhizal fungi (AMF) are ubiquitous in arid ecosystems, yet their distribution and community structure along spatial and ecological gradients remains insufficiently explored at regional scales. Here, we employed Malva sylvestris L., a native spontaneous plant species, to investigate the distribution patterns, phylogenetic structure, and community interactions of AMF and the associated root microbiome in dryland ecosystems. Sampling was conducted along a 700 km transect extending from the Atlantic coast to inland Morocco, encompassing predominantly semi-arid ecosystems. Amplicon sequencing of the LSU rDNA region of roots and soil samples revealed a highly diverse AMF assemblage spanning ten families, including Domikaceae, Diversisporaceae, Entrophosporaceae, Sclerocystaceae, and Septoglomeraceae, while the most frequent taxa belonged to the genera Dominikia, Entrophospora, Funneliformis, and Rhizophagus. Phylogenetic alpha diversity declined with increasing soil phosphorus (P) and nitrogen (N) but increased with soil potassium, precipitation, and distance from the coastline. AMF community dissimilarity in the rhizosphere was primarily explained by distance from the coastline, MAT, and precipitation together with soil P, N, whereas AMF communities in roots were mainly structured by soil P, N, and carbon. Community assembly processes among root-associated AMF were mainly shaped by total soil N and P: total N drove local AMF community structure (positive Nearest Taxon Index) while variation in soil P increased community turnover among locations (positive beta Nearest Taxon Index). Consequently, network topology was negatively correlated with soil P, and temperature, but positively with precipitation. Specialized AMF taxa, particularly Septoglomus and Funneliformis, acted as hubs in the root fungal network, whereas generalists such as Rhizophagus and Entrophospora drove cross-kingdom associations, interacting strongly with Rhizobium, Sphingomonas, and Caulobacter. Overall, this study advances our understanding of AMF ecology in dryland ecosystems and introduces an innovative bioinformatic workflow that provides new opportunities for exploring mycorrhizal diversity and functions.