I'm not sure, but I just wondered about the ethics? Do you have to obtain permission from the university for this type of research? 

Hi Wladimir,

Can you give us some more details about what kind of data you have? Are you saying that you have toxicity curves for your 12 chemicals and mixtures in a range of species (which would give you species sensitivity distributions (SSDs) for each chemical), or only in lobster? What endpoint are you measuring? Lethality (which would be LC50) or something else (growth? reproductive success? etc)

I'm wondering if what you are looking for is a Chemical Toxicity Distribution (CTD), which is similar to an SSD, but instead of measuring the effect of one chemical on many species to calculate an SSD, you measure the effects of many chemicals on one species and calculate a CTD.

See Williams et al 2011 Application of chemical toxicity distributions to ecotoxicology data requirements under REACH. Environmental toxicology and Chemistry 30(8), 1943–1954  DOI: 10.1002/etc.583        

And to answer Andrew: Ethics requirements will depend on the relevant legislation in the country where the work is carried out.

Dear Anne

Thank you for your kind response.

In fact, I have the concentration-response curves of 12 different chemicals (4 herbicides and 7 heavy metals) at 12 different test concentrations. In my study, I am working with three different toxicity models (concentration addition, independent action and the combination of both). The endpoint that I am measuring is the lethality (LC50).

As you mention correctly, I want to measure the  effects of many chemicals on one species in this particular case, spiny lobster. Based on related papers which I read, they determined the Hazardous concentration (HC5) and potential affected fraction for sensitivity distribution.

What I am trying to do is to plot the potentially affected fraction of the specie versus the concentrations of the different chemicals gotten from the three toxicity models in order to know sensitivity values that can potentially affect spiny lobster.

I appreciate the answer of Andrew too.

I hope my reply can help to clarify the question.

Thank you for share Williams study, I will check it out.

Best regards

Well, SSDs are looking at toxic effects of one substance on many species, and tell you which species are very sensitive and which are very tolerant. If you assume the tested species are representative of all species, then you can use the distribution to calculate what percentage of species are likely to be affected by a given concentration of that substance. If you are only looking at one species (lobster), then SSDs are not relevant to your problem.

Have a look at the Williams paper and see if that helps. However, now I've seen more of what you are aiming for, I'm not sure it will be useful. The CTD method is effectively the inverse of the SSD method. It takes the toxicity of all the chemicals on your species and gives you a distribution which can be used to calculate what percentage of chemicals is likely to cause a given effect in that species.

I'm afraid that I'm a bit rusty on mixture calculations. However, can I clarify: Have you actually measured toxicity of mixtures of your chemicals, or are you trying to calculate the mixture toxicities from the single-chemical curves using the three models? Is it the calculations that are causing you problems?

Dear Anne

That means that I can use my toxicity data  in order to calculate the percentage of chemicals that likely cause a given effect in the species?

In response to your question, I already calculated the toxicity of chemical mixtures using CA, IA and CAIA toxicity models, in which I obtained 12 new concentrations for each toxicity model.

Thanks Wladimir. But have you got ethics approval? Your experiment sounds a bit cruel 

Dear Wladimir,

Yes, that is the aim of the Williams method. However, it really only applies to groups of similar chemicals with similar modes of action, so I suspect that you wouldn't get meaningful results from a mixed group of herbicides and metals.

As I said, I'm a bit rusty on mixtures calculations, though I've used CA and IA some time ago. What model are you using for CAIA?

I'm not clear on what you mean by 12 new concentrations for each model. If you have 12 chemicals, then there would be 66 possible pairwise mixtures.

Dear Andrew,

I don't know what the ethics regulations are in South Korea, but in many countries crustaceans are not covered by ethics committees, so there is probably no mechanism for Wladimir to obtain ethics approval.

And yes, toxicity testing may be cruel. The ethics issues then boil down to whether the benefits of having the information obtained, outweigh the costs of the testing to the organisms concerned. For example, if we wish to protect lobsters in the metal-polluted receiving waters for a mine, we need information on how toxic the metals are, alone and in mixtures. We do try wherever possible to use existing data, and alternative methods such as in vitro methods such as testing cell lines etc, or mathematical modelling, but these methods have limitations (as do toxicity testing methods, of course). It is a really long stretch to extrapolate from the toxicity of a chemical to cells in a Petri dish, to deciding what concentration in water will affect a lobster. (Always assuming you can obtain a suitable cell culture from lobsters, of course.) Different tissues have different sensitivities; whole lobsters have many physiological mechanisms for excluding or detoxifying metals; and the nature of the water (amount of organic matter,  pH, oxygen levels etc) can influence metal toxicity. Mathematical modelling is only as reliable as the data that is put into it, and obtaining that data requires testing.

Without reliable toxicity data we have no chance of regulating industrial polluters to minimise harm. In the real world we aren't going to be able to do away with mines and industry, because we need them to survive as a society. And the environment can tolerate and clean up a certain amount of pollution. We need to be able to work out how much, and for that we need toxicity data as well as all the other information about physical and chemical processes influencing pollutants in the environment.

This is an interesting and important topic, but I'm afraid I don't have time to go into it further. (It's nearly 11 pm, and I still have a lot of work to do tonight.)  I suggest that you post it as a new question on Researchgate. Then you will get a much broader audience. I'll follow it with interest!

Regards,

Anne

Andrew,

In case my point got lost in all that text, the ethics question would be: does the cost to the lobsters that are tested, outweigh the benefits of protecting the (hopefully) thousands of lobsters in the receiving waters from death by metal poisoning?

Anne

Dear Anne

Taking into account the assumption of CA model in which the compounds have similar mode of action is possible to apply Williams method.

CAIA is the combination of CA and IA toxicity models. Based on multiple linear regression is possible to develop CAIA toxicity model.

Let me share you Qin et al study (2011). A novel model integrated concentration addition with independent action for the prediction of toxicity of multi-component mixture for your reference.

Originally I have 12 concentrations of 12 different chemicals. Applying the mathematical equations of the CA, IA and CAIA model you will obtain 12 concentrations by each model because the aim is to predict the toxicity as a mixtures not as a single compound.

I already read Williams study and other related papers. I applied a CTD in my log-transformed toxicity data in the form of cumulative distribution in order to  calculate the percentage of chemicals that likely cause a given effect in the species

Dear Wladimir,

Thanks for the information about the Qin paper. It looks like a useful method. 

It sounds as though you have your methods pretty well sorted out. Hope things go well for the project.

Regards,

Anne