Philip Dixon is a professor of statistics at Iowa State University. Dixon research interests include developing and evaluating statistical methods to answer interesting biological questions. Some of his current projects are developing non-parametric estimates of prediction distributions, modeling physical activity data, and developing model-based visualizations of species composition data.
Episode Description
Earth day was launched in 1970 in the aftermath of several environmental disasters in the publication of Rachel Carson Silent Spring. It was designed to help raise awareness of environmental issues and has since grown into a global event. With this year's Earth Day taking out a particular urgency in light of the most recent UN Climate Report. But what goes into the scientific research that informs some this activism? What statistical tools are used to better understand the health of our environment. That's the focus of this episode of staffs and stories with guest Philip Dixon.
+Full Transcript
Rosemary Pennington
Earth Day was launched in 1970. In the aftermath of several environmental disasters and the publication of Rachel Carson's Silent Spring. It was designed to help raise awareness of environmental issues, and has since grown into a global event with this year's Earth Day taking on a particular urgency in light of the most recent UN Climate report. But what goes into the scientific research that informs some of his activism, what statistical tools are used to better understand the health of our environment? That's the focus of this episode of Stats and Stories where we explore the statistics behind the stories and the stories behind the statistics. I'm Rosemary Pennington. Stats and Stories is a production of Miami University's Department of Statistics and media journalism and film, as well as the American Statistical Association. Joining me is regular panelist John Bailer, Chair of Miami statistics department. Our guest today is Philip Dixon, a professor of statistics at Iowa State University. Dixon's research interests include developing and evaluating statistical methods to answer interesting biological questions. Some of his current research projects are developing nonparametric estimates of prediction distributions, modeling physical activity data, and developing model based visualizations of species composition. Data. Phillip, thank you so much for joining us today. It's a delight. Just to get our conversation started. I'm wondering how you got interested in doing research related to biological issues.
Philip Dixon It actually started when I went to graduate school . My PhD is actually in ecology, plant ecology in particular. And my major professor at Cornell, you took two minors along with a major graduate topic. And my major professor said, you will minor in either soils or statistics. And in the Ithaca area, upstate New York, the Finger Lakes, the geomorphology is fantastic. And the geomorphology course was absolutely wonderful. But the second course I took on the taxonomy of soils was the dullest thing. And that's coming from somebody who had an extensive background in plant taxonomy. It is a flat tax for just being boring. And it turns out, I was good at statistics. And after a few years doing the minor, which was a pretty substantial commitment that the minor professor said, why don't you come get a masters, and I had some things that they didn't know how to solve in the biometrics unit. So that became my masters. And when I went on the job market, I had two interviews as a quantitative ecologist in biology departments, and one is the statistician at Savannah River Ecology Lab, advising ecologists, and that was the only job I got offered. So South Carolina we went. And a few years later, the job at Iowa State opened up and it was time to leave Savannah River and it was wonderful to move to Iowa State. And so that's what I've done. And my wife thought she'd be married to a poor ecologist, the rest of her life turned out to be quite a bit better than
John Bailer That's, that's wonderful. I, you know, I'm sort of thinking about, you know, so you thought statistics was better than dirt.
Philip Dixon That is so true.
John Bailer I just feel so affirmed.
Rosemary Pennington Fighting words, though. In Iowa, it's a lot of farmers out there who take their dirt very seriously.
John Bailer I was a little reluctant. I didn't know I would get in trouble for saying that. But it just, it just was bouncing around my head when you said that, Philip. So can you talk a little bit about the Savannah River Ecology Lab? I mean, what kind of stuff was being done there, and what were the ecologists working on that you got involved in early in your career?
Philip Dixon
So anything and everything that was relevant to the ecology in the environment of the Savannah River site, so this is now no longer active, they're now in a reclamation mode. But from the 19, early 1950s, until the early 1980s, mid 1980s, they were the source of most of the nasty stuff that went into bombs, nuclear bombs, and you when they set it up, they kicked everybody off 300 square miles. So that was one of the smartest decisions they made, they found an area where you actually didn't have to move too many people. And you could put your industrial facilities in the middle of a very large buffer area. But that meant that there were all sorts of interesting ecologies going on in both succession and the effects of the nuclear operations, mostly that was hot water, because the reactors were cooled very inefficiently. A little bit of radionuclides stuff because they were accidental, very low level accidental releases, and just what's going on in the southeastern coastal plains. So there were people doing botany, herpetology, soils, you name it. And for a brand new statistician, it was immediate exposure to a huge range of different issues. So the first thing I asked when I got there was that the library got a subscription to a couple of major statistical journals, so I can keep up with what I needed to learn.
John Bailer That's, that's really, that's really interesting. I mean, I hadn't appreciated the buffer aspect of where that facility had been located. Well, among the projects that you worked on there was one in particular that stood out for you as, as kind of a fascinating investigation.
Philip Dixon So a lot of the stuff was mostly local. But one of the things that people don't really realize is they have and they still have going on today, the longest running demographic study, frequent demographic study in the world. So there's some wonderful little features on the southeastern coastal plain cold Carolina bass. These are the southeastern version of the vernal pools in California. So they fill with water, December, January, they hold water through March. And they're major amphibian breeding grounds. So they migrate in breed, and then migrate back to the uplands over the summer. And because one of those was going to be impacted, the lab was funded to set up a study on how many amphibians use it, you know, sort of basic descriptive stuff. In order to do that, what you do is you build an aluminum fence, so the amphibians can't climb over it. And you put buckets every 100 yards or 50 yards around the perimeter, both sides, so newts, you know amphibians of all sorts. I know the sweet genius, but I don't know if the common names move in after the ponds fill. And they hit the fence and they walk laterally and fold in a bucket. And somebody comes by every morning, scoops some out of the bucket, marks them and puts them on the other side. The process repeats in the spring when they're migrating out. And so somebody has been walking the fence every day. I think they're now pushing close to 30 years. Oh, wow. And so you've got a date, you know, they're marked by what year they moved in what year they moved out. And you can estimate how long they live in the uplands and how long it takes before they come back. You can get the demographic you know how many came in, how many went out and how that varied with environmental conditions. And it's been used to inform everything from how much buffer do you need around these when they disappear into the uplands? How far away do they go? The incredible resource all started because the lab was hired to do some descriptive work and somebody took his shine to it and has been continuing the study most of his career.
Rosemary Pennington Wow, I can be honest that when you said Cal Carolina Bay's This is how entrenched I think in stats I am now I immediately thought oh, Carolina Bay's like Bayesian analysis. You've destroyed me, John, you destroyed me.
Philip Dixon If I spelt it out that you'd be of no concern, most people's reaction is well, there's some wonderful coastline. So you know, these are, you know, the various inlets along the coast. Nope. I believe they got their name from the bay trees that are frequently found in them. But I may be drinking and digging up the wrong memory. So that was sort of the way the lab worked. They used stuff that was definitely mission centric, and for the Department of Energy, and figured out ways to make interesting more general science out of it.
John Bailer I'd like to just be before we jump into some of the newer stuff, I mean, well, we never really have talked about and others may not or may not appreciate is it's going to when you talk about ecology, what when you think about the study of ecology, could you just give kind of your you know, your your standard elevator description of of what ecology is, and what are some of the problems that ecologists study.
Philip Dixon
So I do wear, I should preface this by saying I wear two hats, an ecological statistician and an environmental statistician. And I did do environmental statistics SRL but ecology is the study of organisms, the environment they live in, the relationships between the organisms, and increasingly larger scale issues that have to do with nutrients and energy flow through natural and man disturbed situations. Man is a part of ecology just as much as animals are. So my dad used to teach an introductory ecology class. And he would always start by saying most of you, we're here to save the whales. We're going to tell you about the tools that are used to save the whales, not about the policy to save the whales. And that might be one way of thinking about the distinction between ecology as a scientific discipline and ecology in the more popular parlance,
John Bailer
So you've talked about ecology as a scientific discipline, then how do you think about environmental statistics, as it might differ from that, the operational distinction.
Philip Dixon
I draw eyes, whether there's a critter involved, whether it be a plant or an animal. And so when I teach environmental statistics, it's more the sorts of stuff understanding what's in the environment, looking at what affects the environment has toxicology, to me is part of environmental statistics. This is coming near and dear to John's interests. And we did all of those at the Ecology Lab. Mike probably knows my name, Mike Newman, who worked at the interface of toxicology and statistics for many years. He was one of my colleagues, and got me interested in toxicology. Other colleagues there got me interested in below detection limit, which is a long standing battle between analytic chemists and statisticians, which, so far the analytic chemists are winning and still winning.
Rosemary Pennington Well, do you want to explain what that is then for the people who are going to be unfamiliar?
Philip Dixon Sure. So if you want to go imagine some nasty, this could be lead concentrations around a an old building with peeling lead paint in, in an inner city somewhere, it could be radionuclide concentrations, either at a small scale, or around a dump, or of a large scale from, say, Chernobyl, it could be Mercury concentrations in water or in fish. And so you go extract a sample, send it off to the lab, the lab, the Campbell app does some very sophisticated chemistry. And sometimes they'll tell you, there's 10 micrograms per gram tissue, and they give you the number. And other times they put their head in the sand, I'm being a bit biased, and I will confess it. And they will simply tell you, there's less than five, this is a focus on a single sample is the best way I can sort of put my mind in the head of an analytic chemist, where they're reluctant to say there's something there if they're not sure that it's really not zero. And so that's all fine and good if you're focused on a single sample. But when you're charged with understanding the spatial distribution of something, or you want an average concentration in fish, which is going to influence whether or not you close the fishery, you're not interested in the single sample, you're interested in things like the population average, which gets statisticians involved. And it's not immediately obvious how you can put 1015 20, less than five less than five less than five into Excel and calculate a mean, it gets rather mad at you because of the less lands and that's a good thing. So there are a variety of ad hoc solutions, most of which are one step away from scientific lying when you make up a number, but there are other ways of ranging everything from maximum likelihood to nonparametric methods that are used to make statistically justifiable conclusions about these data with less than some limit values. It's the environmental version of survival data where in both cases, it's a very good outcome for survival data, somebody lives longer than your study. For environmental data, it's a very low concentration of something. But you still got to calculate an average or some other descriptive statistics.
Rosemary Pennington You're listening to stats and stories. And today we're talking ecological and environmental stats with Iowa State University's Phillip Dixon. Now Phillip, I know that there have been intersections in your life between the statistics and the soil that you studied. So I wondered if you could talk about a project where stats and stories soil came together for you.
Philip Dixon So when I moved to Iowa State, part of my job was running the consulting, Statistical Consulting operation here, which means I get it's a statistical nirvana. I get to work with incredible bright people asking all sorts of interesting questions. And they often push my statistical limits, which is what I love, figuring out what they're doing and whether they can be done better. And so one of the big things here in Iowa and it has been for a number of years is export of nutrients into creeks and then into the Mississippi River. And then out to the mouth of the Mississippi River and the consequences on the Gulf of Mexico. I mean, I was not the only source, but it's a major one. And one of the projects I was involved in quite early on, was looking at highway embankments and how we can cut erosion off of the highway embankments. The soils in Iowa, at least in the middle part of Iowa, the really agricultural productive area, are very, very fine silt clays. And we get thunderstorms in the summertime. And the combination leads to some rather serious erosion, especially if you've got a slope. And we do when you've got a highway embankment see around an overpass on the freeway. And so the question is, what can we do to reduce that? And that's amenable, that level of a question is amenable to an experimental study? So you go find a site that's willing to let you experiment on it. And in this case, the work I was most heavily involved in was looking at the application of composts. So a variety of different types of compost, some sewage sludge, some other sorts of compost and asking the question, how much do they reduce the erosion when it rains. And so the neat thing in this study was, if you've ever… Have you ever had a garden, you know that using a hose can be rather difficult, because spraying water from a hose isn't really like the nature of rainfall. And so you don't want to just take a big one of these water tankers and just dump water on something, you want to actually make it sort of act like rainfall. And the Ag Engineering department here has built a variety of different rainfall simulators where they hook it up to a big water tank, and they then simulate what a rainfall event would look like with small droplets and big droplets. And they apparently get rather close to what natural rainfall looks like. And so then you just collect the stuff coming out of the bottom of the experimental plot and look at the differences in erosion. And lo and behold, compost actually does a very, very good job because it soaks up, it absorbs the water and stops it from the impact and vastly reduces the erosion.
John Bailer Yeah, I was really intrigued to see that idea of the rainfall simulator being part of the study that was conducted. And so you know, as you just described it, I'm kind of picturing, you know, someone doing this, this very, very purposeful, systematic study of the composition of rainfall, you know, and thinking about how do you basically, you know, deconstruct this storm, to basically build the simulator of it. And that just sounds really cool.
Philip Dixon Yeah, I came in after they'd already built these simulators. So I couldn't tell you much about engineering. Other than that, they went to an awful lot of effort to try to get it to create things that look like natural rainfall patterns.
John Bailer So are there compost being used in Iowa for some of this restoration work?
Philip Dixon
I don't think so. It's too expensive.
John Bailer It's better to receive it multiple times and then to try to set it up. Yeah. Okay. Well, that's an implementation issue.
Philip Dixon I mean,that's right. That's what matters. And how much do you do it cost if you do have erosion going down, but I mean, the study showed very clearly that if you are willing to find a source of compost, it's a good idea.
John Bailer So you've you know, other recent work that you've been involved in is looking at kind of recovery of restored wetlands. I mean, there's a, it seems like there's a common thread that you might see in the work that's done in ecological statistics or environmental statistics. One is, you know, how much is too much when you're thinking about these kinds of impact studies like toxicants, or some other things that are in a system. The other is, is what should assist, what should be the functioning of a system? And how do you move things back to kind of healthy functioning, given impact? So what would you think about kind of understanding like, you've talked about these coastal wetlands? What are things that you've been working on recently? And what are some of the challenges and some of the insights that you've gained from a game from it?
Philip Dixon So that's a lot of questions.
Rosemary Pennington John is trouble for that,
John Bailer You know, holy cow, you don't know. Now Rosemary is gonna give me even more trouble than she does. Sorry, Philip, I am just Okay, so let me see if I can, if you can't deconstruct I might be able to. So you're evaluating impact. That's part of what you do. You're evaluating restoration after impact. How do you know if it's worth the restoration process?
Philip Dixon Hi, I can ask a question. Right? Well, and that's been one of my long standing sort of taking something that's been well developed in one area and figuring out how it applies, and how it needs to be tweaked in some way to to apply it in another area. So the question here is, and you can view it a couple of different ways. You can either ask, can you show that something has no impact? Or alternatively, how do you show when you've restored something, you've restored it back up to where it was? If you're aware of equivalence testing, or bio equivalence, or the generic drug approval process, from human medical statistics, you immediately recognize that question? It's the same sort of question that the FDA has to decide on a daily basis, whether a generic drug is equivalent to the very, very well studied name brand prescription drug that's about to come off patent, you're trying to show this is no difference. And as I tell my students, when I teach this, doing a small, sloppy, horrible measurement error study, and getting a p value, your traditional hypothesis test and the p value of point nine does not show there's no effect. So that leads to turning the hypotheses around, which can be done in a variety of simple or complex ways. And trying to show that you've actually precisely measured something that's in fact, demonstrably small. So that idea can be applied to restoration, asking the question, if you've studied restoration, usually, you get some things that respond very, very quickly and come back to baseline conditions or reference conditions fast. Other times, you aren't there yet. But you can extrapolate and then ask the question, just how long will it take before these are projected to be similar. So the work I was doing on coastal wetlands, a big meta analysis of recovery in coastal wetlands was part of the work that I was involved in with the Deepwater Horizon oil spill, which is what 2011 huge oil spill huge environmental impacts from both the oil and then in some places, and finer, the surfactants used to remove the oil. And so NOAA, through a statistical consulting company hired six statisticians and I was one of them to basically have oversight of the huge amount of work that was being done to understand all the different consequences, vegetation, Marsh, erosions in Louisiana, dolphins, everything all the way down to juvenile fish, tuna, for example, cardiac performance, which apparently makes a difference. I didn't know that people could study the morphology of hearts of juvenile fish, but apparently you can and detect abnormalities. And so you know, it was a huge, wide ranging, any sort of ecological environmental topic that came up sometime during the Deepwater Horizon. It's called the natural resources damage assessment. It's the formal policy when there's an oil spill to assess the consequences of that. I was very glad when that was settled. Because I would have been on a courtroom stand, witness stand testifying for a long time, because I was involved in a whole lot of those assessments.
Rosemary Pennington
You know, I wonder, we don't have a whole lot of time left. But given sort of your career, and the fact that sort of stories around ecological or environmental issues tend to wax and wane. Like, you know, I think right now, there's a lot of stories about the microplastics in our bodies. But then, you know, it'll leave the need to leave the news agenda. At some point, I wonder if given sort of your work and sort of what you've seen if there is a story that you feel like people should be paying closer attention to?
Philip Dixon Well, the one that we should be paying closer attention to is obviously climate change and sea level rise. I'm currently working on an evaluation of the Everglades and progress towards restoration of the Everglades, which is an every two year National Academy of Science Committee. And one of the things that's really apparent down in Florida, south of Miami, when you can be two or three miles from the current ocean, and you're only one and a half feet above sea level. It doesn't take much sea level rise before you start losing an awful lot of land. So that's one that obviously attracts a huge amount of attention, but I'm not sure we're paying enough attention to that attention. But I think you put your finger On the one that is the hidden story, and that is the huge amount of plastics that are currently scuttling around the ocean. I've been out in the Central Pacific on a coral reef out in the middle of nowhere. And the beach was just full of plastic out in the middle of nowhere, and we're not talking about anywhere near the garbage gyre. This was just pick your at all. There's a lot of plastic out there. So I would say that would be my candidate for the most unappreciated story. And we just don't know what those do. They don't, they don't decay very quickly, but they're all out there.
John Bailer
So one thing Philip, we often ask is, you know, if people are interested in, in working in ecological or environmental statistics, or or maybe working with journalists helping to prepare stories about about ecological environmental statistics, what what advice might you give to these folks,
Philip Dixon
It helps to have broad interests. That's one of the things that I think has helped me most is knowing a little bit about an awful lot of stuff. And that's both statistical stuff used on the back behind my office and they saw the bookshelf. That's one way I can keep up with things in a whole lot of different topics is to read widely. But it also helps to have a scientific background. And that's where I think if you're interested in statistics, there may be value in getting an undergraduate degree, for example, in some Applied Science, enough math to be able to understand the theory that underlies the statistics. But then you have a perspective. And that's one of the things that's helped me collaborate with people in agronomy and animal science and natural resources, is they can tell me something. And while I confess, immunology still leaves me cold. I just don't understand it. And from my conversations, I'm not the only one that struggles there. But generally just having a broad background, and initially, a wide range of interests, is what I would recommend for the sorts of things that I do.
Rosemary Pennington
Well, that's all the time we have for this episode of Stats and Stories. Phillip, thank you so much for joining us today.
Philip Dixon
Absolutely. A pleasure.
Rosemary Pennington
Thank you. Stats and Stories is a partnership between Miami University’s Departments of Statistics, and Media, Journalism and Film, and the American Statistical Association. You can follow us on Twitter, Apple podcasts, or other places you can find podcasts. If you’d like to share your thoughts on the program send your email to statsandstories@miamioh.edu or check us out at statsandstories.net, and be sure to listen for future editions of Stats and Stories, where we discuss the statistics behind the stories and the stories behind the statistics.