Microbial interactions dictate community structure and function yet are difficult to identify in a high-throughput and accurate way directly from the environment. Co-occurrence networks can be used to identify potential associations between microorganisms in-situ using modern molecular approaches. Often times, network analyses are carried out using species abundance data that have been collected through time-series sampling efforts; however, the data collected through these time-series projects violate the statistical assumption that the samples are independent of one another. The inherent time-dependence that may be prevalent in time-series data has the potential to confound network predictions. In this article, Samantha Gleich, a Ph.D. student in the Caron Lab and Dr. JL Weissman, a postdoctoral scholar in the Fuhrman Lab demonstrate that seasonal and long-term microbial abundance patterns decrease the predictive power of two commonly employed network analysis methods. Additionally, this study describes and validates a statistical data transformation that can be used to reduce the influence that time-series features have on network analysis results. The full study can be found in ISME Communications.