Hello!

First of all, I'd argue that results for toxicity impact categories should not be compared directly among each other, as they reflect different midpoint impact categories entirely. Many compounds will only have characterization factors (CF) for some of the impact categories, and those that have CFs for more than one, might have disproportionately larger CFs for one (e.g. marine ecotoxicity), compared to for example freshwater ecotoxicity.

Given that these CFs depend on:

1) Dose-response relationship (effective dose causing a disease probability of 50% for a specific intake of the emission).

2) Exposure and Fate (fraction of the mass emitted into a specific compartment and assimilated by the population, by ingestion and inhalation).

The CFs will vary among different categories according to how these two variables are affected, respectively. Also, human toxicity factors many times are extrapolated from those of animal species, thus results for human and animal toxicities, probably should not be directly compared.

I'd suggest you read this thorough manual on the USEtox tool if you want to learn more on the CFs behind toxicity impact categories.

(https://usetox.org/sites/default/files/support-tutorials/USEtox_Manual.pdf). The toxicity model in CML uses the same rationale that USEtox uses.

Moreover, it is also important to note that similar models to USEtox derive CFs for marine and terrestrial ecotoxicity which deviates from consensus in the sense that data for dose-response curves in marine and soil species is more limited. Tipically, there is more data for freshwater species dose-response curves, as well as for humans.

ILCD discusses this a bit more in detail in their 2011 recommendations (https://eplca.jrc.ec.europa.eu/uploads/ILCD-Recommendation-of-methods-for-LCIA-def.pdf).

Hope my answer is helpful!

Kind regards,

Gustavo