The Fort Cobb Watershed in Oklahoma has diverse biogeophysical settings and provides an opportunity to explore the association of water quality with a diverse set of landscapes during both wet (April 2007-December 2009) and dry (January 2005-March 2007) periods. The objective of this work was to identify spatial patterns in phosphorus (P) (soluble reactive P [SRP] and bioavailable P [BAP]) associated with landscape metrics for two distinct streamflow regimes. Spatial autocorrelation of P was evaluated using contiguous (side-by-side) and upstream (upstream:downstream) connectivity matrices. Biogeophysical metrics were compiled for each contributing area, and were partitioned based on association to P concentrations. Results for both SRP and BAP indicated that spatial autocorrelation was present (p < 0.05). There was more spatial autocorrelation and stream P concentrations were three to five times higher in the Wet phase than in the Dry phase (p < 0.05). Analysis with recursive partitioning resulted in higher R2 with spatial autocorrelation than without spatial autocorrelation and indicated that lateral metrics (topography, soil, geology, management) were better predictors for SRP than instream metrics. During Wet phase, lateral metrics indicative of rapid surface and subsurface water movement were associated with higher P stream concentrations. This research demonstrated that we can detect landscapes more vulnerable to P losses and/or contaminations in either drought or very wet periods.