Seth Perlman/AP
I love numbers. A picture may be worth a thousand words, but I think a good bar graph can be worth a thousand pictures.
But three times in the past few days, I've come across statistics in reputable-looking publications that made me stop and say, "Huh?"
I did some investigating so you don't have to. And indeed, the numbers don't quite tell the story that they purport to tell.
So here goes: My skeptical inquiry into statistics on herbicide use, soil erosion, and the production of fruits and nuts.
First, weedkillers (and GMOs). I was struck by this graph, which appeared in a report issued by Food and Water Watch, an environmentalist group.
The report was published a while ago, but Tom Philpott reused it recently in a post for the website of Mother Jones magazine. The point of the chart is to show how increases in herbicide use on soybeans, corn, and cotton have gone hand-in-hand with the rise of genetically modified, herbicide-tolerant, versions of those crops.
That link seems logical, but still, farmers have been planting more corn and soybeans in recent years. How much of this soaring curve is simply because farmers have more acres to cover?
I dived into the USDA numbers, and discovered, first of all, that they're fragmentary. In recent years, the USDA didn't collect such numbers for all three crops in all years. The curve, in this case, is based on just two data points.
No matter. I took the numbers that were available and divided them by the number of acres planted. (I'm using a column chart to make clear for which years we have data.) Suddenly, the trend doesn't seem quite so dramatic.
NPR using USDA data
And how do we know if herbicide-tolerant GMOs are driving this increase? What if it's something else entirely? For comparison, I decided to look at herbicide use in wheat, since no GMO wheat is being planted. Here's a graph of herbicide use in wheat, per acre, over the same period of time.
Whoa. No GMOs here, and herbicide use went up at a faster rate. (In absolute amounts, farmers still use much less herbicide on wheat than on soybeans or corn.)
NPR using USDA data
First, farmers are reducing their use of tillage to control weeds, in part to conserve their soil. Many are relying more on chemical weedkillers instead. Second, with grain prices high, farmers are more inclined to spend more money on anything that will boost yields.
Both of these factors are probably influencing herbicide use in corn and soybeans, too. They may be more important than the popularity of GMOs.
One thing, though, is perfectly clear. The rise of glyphosate-tolerant GMOs did persuade farmers to use much more of that particular chemical. Some argue that a new generation of GMOs that are tolerant to other weedkillers will drive further increases in herbicide use.
Maybe they will. I'll wait for the numbers.
Next up, soil erosion. Here are two maps that caught my attention. They're published in a report called the National Resources Inventory, released last week by the USDA's Natural Resources Conservation Service. (I should also tell you that the NRCS is one of my very favorite federal agencies; please don't hold this post against it.)
The dramatic shrinkage of those red and orange blotches along America's major rivers is terrific news. It shows that less topsoil is washing away today, compared with 1982.
Cox already knew about this map. He wasn't happy about it. In his view, it obscures more than it reveals.
According to Cox, the good news is old news. Practically all of the dramatic progress in fighting soil erosion occurred 15 years ago, between 1982 and 1997. At that time, "we were on a really solid pathway to finally getting on top of this ancient enemy," he says.
Since 1997, however, progress has stalled, so the map paints an overly cheerful picture. (In fairness to NRCS, there is another, less prominent, graph in the report that does show this stagnation in anti-erosion efforts.)
In addition, there's a basic problem with these data, Cox says: "They only capture one kind of erosion," called sheet and rill erosion. This is the erosion that happens evenly across a field, and can be predicted from the amount of rain, the field's slope, its soil type and whether it is bare or covered by grass. The NRCS data are based on such predictions, and the estimated improvements since 1982 happened mainly because farmers are tilling less, and protecting more of their land with vegetation.
By contrast, no models can predict when something more catastrophic will occur; when small rivulets of water combine into larger, fast-moving streams that cut deep ditches, or gullies, into a field. According to Cox, those gullies actually carry off more soil than the predictable kinds of erosion, and they were especially bad during the storms that hit the Midwest last spring and summer.
So my straightforward good-news story about soil erosion evaporated.
Finally, there was a second surprising statistic buried deep inside that NRCS report, and I noticed it only because of a press release that the USDA put out. According to that release, the NRCS's National Resources Inventory detected "a boom in land dedicated to growing fruits, nuts and flowers, increasing from 124,800 acres in 2007 to 273,800 in 2010."
Wow! I looked at the numbers again. In fact, the boom was only in a category of production called "cultivated" fruit and nut production. Turns out, that's a tiny category, barely worth counting. It apparently refers to orchards in which there's also some tillage going on to grow a second crop.
"Uncultivated" fruit, nut, and flower production, by contrast, went from 4.7 million acres in 2007 to 4.4 million acres in 2010.
Sorry. No boom.