Bipartite structural evaluation: Extended network generation model and corrected randomization techniques

Understanding the structural organization of bipartite networks is essential for analyzing complex systems across diverse domains, including ecological communities. Such networks often display distinct architectural patterns--particularly nestedness and modularity--that influence system stability, diversity, and dynamics. Despite extensive research, accurately replicating these structures remains challenging, partly due to limitations in existing null models and synthetic network generation methods. Traditionally, nested and modular network ensembles are derived either from ad hoc constructions or large sets of random networks, with structural patterns evaluated a posteriori. In this article, we present a comprehensive framework that integrates advanced null models with a synthetic network generator to rigorously assess bipartite network architectures. We evaluate the statistical significance of nestedness, modularity, and in-block nestedness scores using five null models across approximately 25,000 synthetic bipartite networks. Our analysis identifies systematic biases in certain models and introduces a Corrected Probabilistic model to address these issues. By combining the analytical simplicity of our network generator with a robust null model, our framework enhances the analysis of bipartite networks, offering valuable tools for ecology and beyond.