Integrating Bioavailability of Metals in Fish Population Models

Abstract

Population models are increasingly being used to extrapolate individual-level effects of chemicals, including metals, to population-level effects. For metals, it is also important to take into account their bioavailability to correctly predict metal toxicity in natural waters. However, to our knowledge, no models exist that integrate metal bioavailability into population modeling. Therefore, our main aims were to 1) incorporate the bioavailability of copper (Cu) and zinc (Zn) into an individual-based model (IBM) of rainbow trout (Oncorhynchus mykiss), and 2) predict how survival-time concentration data translate to population-level effects. For each test water, reduced versions of the general unified threshold model of survival (GUTS-RED) were calibrated using the complete survival-time concentration data. The GUTS-RED individual tolerance (IT) showed the best fit in the different test waters. Little variation between the different test waters was found for 2 GUTS-RED-IT parameters. The GUTS-RED-IT parameter "median of distribution of thresholds" (m w ) showed a strong positive relation with the Ca 2+ , Mg 2+ , Na + , and H + ion activities. Therefore, m w formed the base of the calibrated GUTS bioavailability model (GUTS-BLM), which predicted 30-d x% lethal concentration (LCx) values within a 2-fold error. The GUTS-BLM was combined with an IBM, inSTREAM-Gen, into a GUTS-BLM-IBM. Assuming that juvenile survival was the only effect of Cu and Zn exposure, population-level effect concentrations were predicted to be 1.3 to 6.2 times higher than 30-d laboratory LCx values, with the larger differences being associated with higher interindividual variation of metal sensitivity. The proposed GUTS-BLM-IBM model can provide insight into metal bioavailability and effects at the population level and could be further improved by incorporating sublethal effects of Cu and Zn.

Extrapolation of Metal Toxicity Data for the Rotifer Brachionus calyciflorus Using an Individual‐Based Population Model

“…Water chemistry was not reported for the population experiment used in our study for validation (Janssen, 1992) and important parameters such as dissolved organic carbon are unknown. For the aims of the present study, bioavailability modeling of metals was not required but the possibility of bioavailability correction of metal toxicity in population models has been demonstrated for rainbow trout (Janssen et al, 2021). Tools specific to metal risk assessment can thus be integrated in ecological models and enhance the applications of ecological models for metal risk assessment.…”

“…Ecological modeling, and population modeling in particular, is an alternative to this approach that utilizes the available toxicity data to allow a mechanistic extrapolation to the desired protection goals (Galic et al, 2010;Larras et al, 2022). Practical applications of population models have focused mainly on plant protection products (Dohmen et al, 2015;Galic et al, 2012;Garber et al, 2022) but examples in a metal context are also available (Janssen et al, 2021;Vlaeminck et al, 2019Vlaeminck et al, , 2020Weighman et al, 2023).…”

Modeling Full Life‐Cycle Effects of Copper on Brook Trout (Salvelinus fontinalis) Populations

“…To implement the IT model, we assumed that the threshold for both adult and juvenile brook trout was reached within 30 days of exposure, which is supported by model fits of Cu toxicity to the related species rainbow trout (Janssen et al, 2021). We implemented the IT model by killing a fixed fraction of randomly selected adults or juveniles after 30 days of exposure, corresponding to the observed relative survival fraction at the end of the test (Supporting Information, Table S1-5).…”

“…Because the chosen method of implementation for survival effects (IT or SD) can affect population-level predictions, this led to additional uncertainty in our population-level predictions. In an earlier study, Janssen et al (2021) found that the IT model was the most plausible model to explain Cu mortality to juvenile rainbow trout, based on results obtained with waters that had a wide variety of physicochemical compositions. Because both our present study and that of Janssen et al (2021) are about trout species, one could assume that the NOEC pop values obtained with the IT simulations may be the most realistic ones in our present study.…”

“…In an earlier study, Janssen et al (2021) found that the IT model was the most plausible model to explain Cu mortality to juvenile rainbow trout, based on results obtained with waters that had a wide variety of physicochemical compositions. Because both our present study and that of Janssen et al (2021) are about trout species, one could assume that the NOEC pop values obtained with the IT simulations may be the most realistic ones in our present study. On the other hand, Gao et al (2017) found the SD model to fit the survival-time concentration data of Cu toxicity to zebra fish larvae the best.…”