Soil Microbial Co-Occurrence Networks Across Climate and Land Use Gradient in Southern Italy

Abstract

Despite extensive research on microbiota across land use gradients, it remains unclear if microbial co-occurrence relationships exhibit consistent patterns. Here, we assessed microbial co-occurrence networks of seven natural ecosystems-Quercus ilex forest, Fagus sylvatica forest, Abies alba forest, Mediterranean and mountain grasslands, and subalpine and Mediterranean shrublands-and five agroecosystems, including vineyards, horticulture, greenhouse, a polluted agricultural system, and an arid greenhouse. Soil chemistry, such as pH, organic carbon and total nitrogen, was characterised, and soil microbiota were profiled using high-throughput sequencing from 242 soil samples. Our results revealed that mountain grasslands had the highest organic carbon (86.4 g/kg), while the arid greenhouse had the lowest (6.1 g/kg). Mediterranean grasslands had the lowest pH of 5.79, and vineyards had the highest electrical conductivity of 0.901 dS/m. Notably, natural ecosystem networks exhibited greater modularity, with protected horticulture showing exceptionally the highest (0.937), while intensive agriculture within agroecosystems had a significantly lower modularity of 0.282. Modularity and the number of modules were positively correlated with soil P₂O₅, while network diameter, path length and clustering coefficient were correlated with soil pH. Additionally, edges and nodes number, average degree and microbial diversity were positively associated with organic carbon and total nitrogen. These findings highlight that natural ecosystems foster more complex and resilient microbial networks, underscoring sustainable land management's importance to preserve soil health and microbial diversity.

Keywords: agroecosystems; microbiota; natural ecosystems; network modularity; next‐generation sequencing; soil chemistry.

FIGURE 1

Images of selected ecosystems in the Campania Region (Southern Italy), showcasing a gradient of climate and land use intensity. The gradient includes variations in organic amendment input, synthetic fertilisers, and pesticide application. All photos by Giuliano Bonanomi and Mohamed Idbella.

FIGURE 2

Co‐occurrence network analysis illustrating potential interactions between bacterial and fungal families across 12 ecosystems. Connections indicate strong (Spearman's ρ > 0.6 or ρ < −0.6) and significant (p‐value < 0.01) correlations. Node size is proportional to the weight of the connecting edges. Nodes are coloured by modularity level, representing sub‐ecosystems within each ecosystem. Module colours are independent across different ecosystems.

FIGURE 3

(A) Bar plot showing the topological average degree (average number of edges per node) for the networks of 12 ecosystems, with lines depicting the average path length (average number of edges in the shortest paths between all node pairs), modularity (extent to which the network can be divided into subgroups with stronger interactions), and average clustering coefficient (degree to which nodes tend to cluster together). (B) Stacked bar plot displaying the relative abundance of bacterial and fungal phyla within the networks. (C) Bar plot illustrating the number of nodes and edges in each ecosystem network. (D) Principal component analysis (PCA) showing the correlation between network topological parameters and soil pH, total nitrogen, organic carbon, and phosphorus.