Do biotransformation data from laboratory experiments reflect micropollutant degradation in a large river basin?

Benchmarking the Persistence of Active Pharmaceutical Ingredients in River Systems

“…ATE was properly identified as fast-degrading . TRI showed consistently strong degradation in the Rhine, which aligns with observed behavior in contact with natural sediments and under riverbank filtration. , LEV degraded in all Rhine campaigns, agreeing with the results of Seller et al RAN degraded fast in the SMPC data sets, in line with the observation of Desiante et al…”

“…Dissipation half-lives obtained by the benchmarking approach showed little agreement with the values obtained by Seller et al from SMPC P1 data (Figure ). The possible reason for this was the noticeable uncertainty in the half-lives of both approaches, caused by the low dissipation signal in the SMPC data and the attached high uncertainty of the measurement points themselves.…”

“…PRO requires microbial adaptation before biotransformation started in laboratory experiments, , and this process could have taken place in the Rhine as it degraded at moderate to fast rates. HYD is phototransformed in the environment with moderate rates and undergoes abiotic hydrolysis too . Out of these two mechanisms, the latter is more likely to explain the observed moderate dissipation rates because phototransformation would be more intense in SMPC P3 than in spring campaigns.…”

“…The possible reason for this was the noticeable uncertainty in the half-lives of both approaches, caused by the low dissipation signal in the SMPC data and the attached high uncertainty of the measurement points themselves. In contrast to this, a clear, almost 1:1 relationship emerged in SMPC P3 (Figure , f (DT 50,model ) = 1.01 × f (DT 50,bench ), adjusted R 2 = 0.84, p = 1 × 10 –9 , for compounds with strictly positive half-lives, where DT 50,bench is the half-life estimated from benchmarking, DT 50,model is the half-life calculated from the Rhine model of Seller et al, and f () in the regression equation denotes a log-shift transformation that maps all DT 50 values to the positive domain: f ( x ) = ln( x ) + 2). The better performance of SMPC P3 compared to SMPC P1 can be attributed to the shorter half-lives and the much lower uncertainty of that data set, which improved the fit of both the benchmark method and the stream network transport model.…”

“…As lab-to-field comparisons are known to be compromised by the dissimilarities between their boundary conditions, comparing indicators derived by different methods from the same data set is more likely to be meaningful. The SMPC data set itself was modeled by Seller et al . using an extended version of the stream network transport and degradation model of Honti et al Model improvements covered the inclusion of hydrolysis and phototransformation, rate constants for which were derived from dedicated laboratory experiments .…”