Anthropogenic beach litter presents a significant environmental challenge, particularly in the north-east Atlantic. Understanding temporal variations is crucial for identifying pollution sources and developing effective policies. However, in-situ data from beach surveys are irregular, both spatially and temporally, and subject to high variability, complicating robust statistical conclusions. This study employs a Bayesian machine learning framework to investigate seasonal variations, identify regional hotspots and elucidate their anthropogenic drivers. Using data from 3,866 surveys across 168 beaches, we leverage a spatial log-Gaussian Cox Process to enhance statistical inference by incorporating information from nearby beaches. Our analysis reveals distinct seasonal patterns, with winter and spring exhibiting the highest pollution levels. Significant regional variances are observed, with pronounced seasonal hotspots along the western Iberian Peninsula, French coastline, Irish Sea, and Skagerrak region. Further investigation into the sources of this litter links riverine emissions and aquaculture activities to the observed seasonal peaks, highlighting their notable impact on beach pollution. This study underscores the need for enhanced monitoring and targeted management strategies to mitigate the environmental impact of macroplastic pollution. Our findings advocate for seasonal monitoring to pinpoint and effectively manage litter hotspots, emphasizing the importance of addressing aquaculture-related plastic emissions.