Cause of residual within-site PMMoV variability

Determine the cause(s) of the residual within-site variance in log10 pepper mild mottle virus (PMMoV) concentration measurements (gene copies per gram dry weight) that remains unexplained by the variance decomposition linear model incorporating latitude, longitude, precipitation, sewer system type, site indicator, and weekly and annual Fourier time components, and ascertain whether microscale variation or measurement noise account for this unexplained variability.

Background

The study partitions variance in log10 PMMoV concentrations across 160 U.S. wastewater treatment plant sites using a linear model with spatial (latitude, longitude), site membership, precipitation, sewer system type, and temporal (weekly and yearly Fourier basis) covariates. Location and site membership explain the majority of variance (~36% and ~24%, respectively), while precipitation, sewer type, and temporal components collectively explain a small fraction (~1%).

A substantial portion of variance remains within sites and is not accounted for by the included covariates. The authors note that this residual variability may stem from microscale variation or noise but were unable to investigate further due to data limitations, leaving the specific drivers of this within-site residual variance unresolved.