Response to L. Witting: PCBs still a major risk for global killer whale populations Our recent Science Report, Predicting global killer whale (Orcinus orca) population collapse from PCB pollution (Desforges et al., 2018) has received much scientific and media attention. This rebuttal by L. Witting is part of ongoing discussions on the online (eLetter) forum of the journal Science, which we acknowledge as useful contributions to the debate on the complex role of PCBs and their effects on killer whale populations. We presently address the specific criticisms by L. Witting as well as emphasize the importance of stepping back and considering again the broad purpose of our study.Given the almost complete lack of knowledge of PCB effects at the population-level in killer whales, coupled with the extremely high blubber concentrations of these chemicals and a wide body of empirical evidence of their toxicity to mammals, our goal was to advance our understanding of the potential risk of PCB exposure in killer whales. Given the lack of information on killer whale demographics, temporal PCB exposure, and species-specific toxicity data, we chose an alternative perspective using a theoretical approach to model relative PCB effects in a "generic" killer whale population exposed to incrementally increasing concentrations of PCBs, and over a nonspecific 100 year time window. This framework was never meant to predict realistic population-specific dynamics or down-play the importance of other stressors such as food limitation, human disturbance, and inbreeding (Ward, Holmes, & Balcomb, 2009;Ford, Ellis, Olesiuk, & Balcomb, 2010;Beck et al., 2014;Esteban et al., 2016), but rather to highlight the often overlooked and potentially serious risk that PCBs pose to a long-lived marine top predator like the killer whale. While we acknowledge that these nuances may have been lost in the condensed text of our recent Science Report, they are clearly stated in the more comprehensive text of the supplemental information.We certainly acknowledged the limitations of our model approach in the main part and supplementary materials for Desforges et al. (2018), and these are further discussed below. Nonetheless, we still feel very strongly that having an empirical evidence-driven risk assessment based on the best available data is a much more preferable approach to a complex issue, than having no assessment at all. In other words, how can policy makers make informed decisions about conservation threats to killer whales when the various threats have not been investigated using the best available scientific and risk assessment tools? We believe our paper achieved this goal, and we now embrace the wider discussion of the topic as it clearly demonstrates the widespread public and scientific interest in the conservation of this flagship marine species.Several comments by L. Witting focus on the key assumptions and parameters of our model design. To be clear, all models have assumptions, make simplifications, and are based on data of varied quality, and o...