Vox points me today to a new paper suggesting the following sequence of events:
- In 2006, a fungal disease called white nose syndrome starts killing off bats.
- Bats fail to eat insects.
- Farmers have to use more pesticide.
- Pesticides kill babies.
- Infant mortality goes up.
Here's the inevitable chart:
The researcher, Eyal Frank, found that insecticide use went up nearly a third—from 33 grams per acre to 44 grams per acre—in counties with outbreaks of white nose syndrome. In those same counties, at the same time as the surge in insecticides, infant mortality went up about 8%—from 6.8 per thousand to 7.4 per thousand—compared to counties that didn't increase insecticide use.
Correlation by itself is not causation, but in this case there's a pretty obvious causal mechanism available. There's probably something to this.
I read the article in the Times. As white nose syndrome spread, it provided a natural experiment. Researchers tracked infant mortality and found it increased following the spread of disease and increased use of pesticides. So it's a pretty strong correlation. And as every good scientist says, more research is needed.
"...There's probably something to this."
I doubt it.
I suppose it depends on the insecticides used (I used to work in the pesticide industry). They vary wildly as to human toxicity. Though there is a caveat: We would be talking about babies, not adult humans who are victims. I don't think they test on any baby animals and if there is special toxicity for young organisms (apart from the fact that children respond to lower doses due to lower body mass) it might be missed.
Having said that: This may be as much caused by careless applicators as by evaporation from the crops, distribution in the air and breathing in by the babies. Something like Dad spraying, neglecting wearing protective equipment (which is not comfortable) and hugging his children right afterwards. Or: children playing near freshly sprayed field.
Declining crop yields also produced declining income, which is itself correlated with infant mortality. I'm also a little troubled by the idea that pesticide exposure causes an easily detectable effect on infant mortality but no other effect on infant health. That doesn't seem very plausible biologically.
Which is not to say that the conclusion is wrong, just that I want more evidence.
Hence my comment. The correlation seems good enough to follow up on and ask those questions. I don't think it gets to the "probable" level.
Green jelly beans cause acne.
https://xkcd.com/882/
One of my all-time favorite xkcd's! I was even thinking of it in a different context earlier today.
I got a bad feeling this research is going to turn out to be guano.
Looks like infant mortality was already increasing before WNS detection.
Lead ---> crime rates go up.
No Lead ----> crime rates go down.
Insecticide(s) ---> infant mortality goes up.
......... we need to see the alternative here.
Why can't the answer be that there were more insect borne diseases?
The insects targeted by these pesticides do not carry human diseases. They cause damage to plants which results in reduced yields.
Plus the insecticides used are chosen to be effective on the crop-damaging species and are sprayed at the optimal time and location for crop protection.
Which makes it rather unlikely that insect borne diseases are the cause.
I was just going to comment on the oddity of the headline which seems to suggest that if the mold was asked it would say I was gunning for the bats, but I did not mean to kill the babies. In this case the consequences seem neither intended nor unintended.
But I see in the comments some of the tendency to take the "correlation doesn't mean causation" line which is meant to suggest a scientific critique but seems to have warped into a anti-Scientific one. I don't mean the people asking for more evidence before drawing a conclusion, which is quite reasonable. I am talking about the people who link to examples which are based on a ridiculously small number of data points, or the people who somehow missed that the evidence above does in fact compare counties with a fungal outbreak with ones that didn't have a fungal outbreak.
One certainly hopes that the analysis was not as blunt as it has to be presented in a blog post. But it is equally silly to simply assume that the scientists did not do a serious analysis of the data. According to other scientists unaffiliated with the study, they did. Of course other scientists can be wrong as well. But at some point it is the people who are just pointing to examples of silly, but weak, correlations who are being anti-scientific, not the people pointing to correlations as evidence of causation.
To just add a little correction to this: There are lots of studies of this kind that can not be reproduced. There are even more that nobody bothers to try and reproduce.
So to quote the standard line about correlation is mostly justified.
In this particular case there is at least a (semi-)probable mechanism describing how the correlation (if real) works which helps to strengthen the argument. But even so one should be careful with assigning blame.
Too bad the fungus didn't kill the bat that passed on sars2