Where do research ideas come from?
Some graduate students have little chance to develop research ideas – the grants that support them already have the ideas worked out. Others have more latitude for coming up with ideas. But how do scientists (esp. students) find ideas to develop? Sometimes examples and case studies are useful in developing insights. At the risk of presenting a boring biography, I summarized some research projects my students and I have done. After typing all this out – some themes seem to recur. The ideas for much of the research my students and I have done depended on:
1. Finding opportune experiments set up by other people, and having a novel idea to test.
2. Identifying an interesting subject that seemed worth learning about – and developing a research project as a way to learn.
3. Developing ideas based on individual papers from the literature – setting out to test them in a new setting, or designing a way that will falsify an unappealing idea.
4. Developing ideas that came from conversations with other scientists, or from presentations at meetings.
Some of the projects actually involved testing real hypotheses (“the lignin:N of fresh litter is related to net N mineralization,” “The flow of alder-fixed N can be traced through the soil and into Douglas-fir trees,”). Many studies focused on finding out how big numbers are (“how does N mineralization differ between species,” or “how much of the C in this soil came from trees or cane?”). The projects that tested hypotheses were a tad more intellectually fun – but the descriptive studies have been very enjoyable too.
Dan’s MS projects
When I arrived in Vancouver, I called my advisor, who was leaving town, and he asked if any idea from a list he sent me sounded appealing – I said the idea of herbiciding an alder/conifer stand and then looking at the water chemistry in the soil and stream sounded interesting – and he said, “Ok, that’s yours.” So I spent more than a year building lysimeters, installing them, sampling them (in a control stand, and a stand I planned to spray) – and had begun lining up the helicopter to spray the site, when August turned out to be incredibly dry, the alders lost their leaves, and I was out of luck (the herbicide wouldn’t work). Throughout this period, my advisor had never come to the field site – I’d have been lost without help from other grad students. I came up with a backup plan – comparing my water chemistry profiles for the 18 yr old mixed stand, with previously measured profiles for a 180 yr old conifer forest just uphill – putting it in the context of succession. No true replication, but the overall story (lots of nitrate where alder is present, not much where alder is absent) is probably robust.
Also did a project where we put Christmas trees over rainfall collectors to look at the extra nutrient inputs that might come with deposition that’s not measured in normal collectors – did it at 2 elevations. The idea came from a similar project that Bill Schlesinger had done as an undergrad at Dartmouth, up on a mountain with great rime ice deposition onto trees.
Kimmins had a consulting contract to develop a bibliography, and a set of data tables, about temperate forest nutrient cycling (as a background resource for his nutrient simulation model) – and he hired me and another student to do much of the literature work. Good experience, good $.
Dan’s PhD Projects
I had visited Oregon at the invitation of a watershed scientist who had visited BC; met Kermit; Kermit had $ to work on N fixation by Ceanothus shrubs. In the final spring of my MS, my advisor happened to be talking to a researcher at one of the timber companies in BC (MacMillan Bloedel), and found out they wanted someone to do some N fixation work; my advisor knew I planned that for my PhD, and told me to call the MB guy. I got a summer job summer doing research designed mostly by the MB guy:
1. Drive around Vancouver Island and map the distribution of N-fixing species.
2. Figure out how much N is fixed by shrubby Sitka alder, using acetylene reduction.
3. Take a site near town where both red alder and Sitka alder occur, and measure their N fixation rates, and the growth of the D-fir in the plantation.
After the summer, I moved to Oregon, and MB began looking for a PhD student to follow up on the work I started on Vancouver Island. I developed a plan to do an N cycling study in a chronosequence of Ceanothus sites – but was disappointed when I finally read all the literature and found Mike Newton had already done that… By February, MB had not found a grad student, and it occurred to me that I might be able do enough on the Ceanothus to meet Kermit’s needs, and do most of my dissertation work on Vancouver Island. So I: took off the spring quarter, designed 2 chronosequence samplings for total soil N under Ceanothus; did that work in April/early May; and headed to BC to launch my dissertation work.
The dissertation had no hypotheses, and had no overall proposal. I wrote vague workplans for the company, but mostly figured I’d take the site with red and Sitka alders, and measure the nutrient cycling features of the stands. The plans eventually grew to include production and biomass work at an alder/D-fir site in Washington of the same age but much richer soil (this idea came from reading the literature, where an author mentioned the existence of this stand), and then using some long-term data for 2 other sites to develop an age x site fertility story.
Along the way, these other research projects developed:
1. Resin bags. Kermit mentioned a seminar by Susan Riha; she had put soils in PVC tubes as mini lysimeters, with a layer of anion resin at the bottom to catch nitrate that leached out. I’d never heard of resins – found out more about them; came up with the idea of burying them loose in a bag in the soil as an index or N supply. Couldn’t think of a good material for making the bag, and inspiration struck while looking at Robin Graham’s legs one day. Designed a trial in my BC stands; almost gave up when ammonium analyses didn’t work with an ion electrode; Pam Matson got the analyses to work on the autoanalyzer; had a bioassay of D-fir seedlings for testing the method…
2. Natural abundance of 15N – Warwick Sylvester gave a seminar course on stable isotope methods, and in a discussion with Bruce Caldwell, the idea came up for using this approach to determine how much of a Douglas-fir’s N came from alder N fixation. MacMillan Bloedel agreed to fund the work; Bill McGill at Univ. Alberta offered to let me run his mass spec (natural-abundance precision machines were not common then!). The results were promising… Phil Sollins helped me write an NSF grant to repeat the work at 3 other sites – this paid for my first summer’s wages after my PhD was completed. Probably the lowest-hassled proposal I’ve had funded from NSF!
3. Leaf area projects: Dick Waring was big on using stem sapwood area to estimate leaf area, and I got several publications from using this technique for looking at relationships between LAI and growth with D-fir spacing, fertlization, and alders; included some testing of sapwood:LAI relationships, but not thorough.
4. Moss – Mosses were “big” in forest ecology circles in BC, but not in Oregon. The most-intensively studied old-growth D-fir forest at the HJ Andrews had an abundant moss carpet, but nobody had measured it. A friend and I spent a day sampling, a day weighing and piecing apart annual growth segments, had some analyses run, and published it in Ecology.
5. Misc.: a self-thinning paper came out of a silviculture class – a new idea to apply to some data I had in hand. Another project looked at decomposition and N supply in clearcut vs. intact forests – just a matter of curiosity (most expectations were that it’d be higher after cutting, but not much data) and easy to test while zooming around on Vancouver Island.
The Duke years:
1. H+ budgets of alder/conifer ecosystems. Phil Sollins had done some pioneering thinking about H+ budgets, missed a key point or two (which Vitousek told me about in a hallway, but I didn’t understand at the time)… Other studies had shown alders acidify soils, and I thought it would be interesting to develop a whole H+ budget to explain why. We had two 50-yr-old sites with conifers and alder/conifers, and sold it to NSF on the second try. The project was never completed as it should have been – ie, we never got to calculate H+ budgets because some key pieces of the budget were not reliably sampled… but we did get 3 pubs from it, including one that answered the “why a lower pH under alder?” question in a way we hadn’t foreseen. I also recall a conversation on a field trip, where I asked Dan Richter and Henry Gholz why forest floors area acidic – and we didn’t really have a good, process level answer to this simple question, so our curiosity developed more. Good sites + good budget = unplanned insights.
2. EPRI Integrated Forest Study. Dale Johnson and Steve Lindberg saw a need for a coordinated study of a variety of forests to simply quantify acid deposition and nutrient cycles. They thought the current state of the science was too fractured and incomplete. They talked EPRI into $12 million, and invited Ken Knoerr (microenvironment guy Duke) and me to install and manage a site with their protocols.
3. H+ budgets on a topographic sequence in Alaska – Dave Valentine’s PhD thesis. Dave was funded from the EPRI IFS project, and had to do something biogeochemical that was relevant to acid deposition. I said he should develop some ideas and then chose the best place to test the ideas (rather than limit himself to a Duke Forest site) – so he decided to do H+ budgets at the Tanana River chronosequence in Alaska. Keith Van Cleve gave us a tour, but said Dave couldn’t work there if he wasn’t a U. AK student (Dave now has Keith’s old position and lab!!). Meanwhile, Knute Nadelhoffer had given me a tour of their sites near Toolik Lake, esp. a topographic sequence they were studying, and Dave developed a different idea about topographic control of H+ and acidity. Once again the H+ was never calculated, because of an unforeseen and unrecoverable problem with the lysimeter solutions – but he had great sites, a good budget, and enough good stuff to graduate anyway.
4. Why soil pH declined over time in a pine plantation – Riding back from a field trip with Carol Wells, he mentioned he had been archiving soils every 5 yr from a post-cotton pine stand in S. Carolina, and that the site had acidified pretty dramatically in 25 yr. He offered the soils if I wanted test any ideas with them. So we used a new approach that Valentine and I had developed – using titration curves to explain why soil pH differs between samples (more acid? More dissociated? Stronger weak acids present?). I developed the general idea from the Henderson-Hasselbalch equation; Dave figured out how we could do it for titration curves rather than monoprotic acids.
5. Prescribed fire effects – heard about a replicated 30 yr study on prescribed fire, with fires every 0, 1, 2, 3, or 4 yr; developed a soil N study for an MS student; came back after Hurricane Hugo destroyed the stands for a final sampling. Good site + slush fund = good science.
6. N availability in spruce/fir forests: Funds were available for work on high-elevation spruce/fir forests, and Carol Wells at the USFS wanted to know what N mineralization rates would be, as well as denitrification rates. He got the funding and shared it with me – and we stretched it to cover the main study (more than a dozen sites for in-field incubations), and some ancillary ones (mineralization in fir waves in NC and NY, mineralization inside and outside wild boar exclosures in the Smokies, and some 15N work in fertilized soils in Washington),
7. Further development of resin bags: Dick Waring volunteered to pay for time in the Phytotron at Duke, and I designed a pot experiment to look at sources of variation in capture of N by resin bags. Treatments were: 2 soils (one had been fertilized 6 months prior), sterilization with methyl bromide (to shut off microbial uptake), addition of cellulose (increase microbial uptake), a range of ammonium nitrate additions, a range of water additions, and with and without plants. Peter Vitousek suggested I include some labeled N on the resins (to test for the need for a biocide to prevent microbes from removing N from resins), gave me some 15N, and told me how to use it.
At an ESA meeting, John Pastor presented some Blackhawk Island work, and invited people who might have ideas to test to come to Wisconsin. Given all the work they already had on buried bag incubatioins, it was easy to see a good chance to see how resin bags compared to buried bags.
Steve Hart developed his MS thesis to look at ways of estimating nutrient availability that would predict tree response to fertilization. Weyerhaeuser supplied the sites and the ticks, Kermit Cromack donated the resins, and Steve funded much of it himself.
8. Species effects on soils in Connecticut: I read a paper by Dave Challinor, dealing with a replicated species trial during his PhD days at Yale in the early ‘60s. I contacted Challinor, he referred me to D.M. Smith at Yale, who told me Challinor’s plots were destroyed by some beetles and diseases – but he knew of some others that might interest me. On a visit to my mother-in-law in New Haven, I visited Smith and saw the sites, and then designed a study to look at species effects on acidity and N mineralization. I planned to have Dave Valentine look after the study, but he planned to have me look after the study, and we ended up losing some of the samples, and the study was a bit of an orphan – but the results turned out great, one of my favorite studies! Hurricane Bob later came along and prevented anyone from double-checking our work.
9. The manager of the Duke Forest planned to burn the understory of a pine stand, as a way to knock back the hardwoods before the pine was harvested -- with the idea that the replanted pines would not have a severe competition problem with the hardwoods. I had worked on a prescribed fire project with Wally Covington at NAU, and the part that always fascinated me was that the fire removed a modest amount of the N in the forest floor -- but over the next year the N in the forest floor declined much more. So a Peter Schoch (grad student) and I quickly sampled the forest floor before the fire, the day after the fire, and 6 months later -- and found the same large N loss from the forest floor through decomposition.
10. Some other student projects:
Kathy Moser, Eric Waldbauer – looked at growth, soil N min., and understory vegetation in a loblolly pine plantation with and without understory legumes – an easy idea given some 8 yr old plots established by Jacques Jorgensen of the USFS.
Craig Hedman – looked at height profiles of canopy leaf area in a range of hardwood stands (across a fertility gradient) in Duke Forest. The idea came from a visit with John Aber, and we copied some of John’s work.
Randy Bell – used the 30-yr prescribed fire plots to use 15N techniques in N mineralization – did fire increase net N min. by reducing the C supply to microbes (immobilization), or by altering gross N min? Turned out immobilization was so rapid (>95% in a day) that we weren’t comfortable using Kirkham & Bartholomew equations.
Qiuyen Chen – no $ available; just learning about redox effects on pH – decided to compare soils in a cypress plantation (swampy) with adjacent, upland pine. Mostly descriptive – turned out difference in pH related exactly to difference in redox – nifty!
Deb Grant – looked at rates of free-living N fixation across a range of stand ages and soil supplements (nutrients, water). Straightforward – no big ideas – just a desire to see how big some numbers are, and what they may be sensitive too; funded by a small grant from Duke for research development.
Beth Anne France – looked at soil acidity under different species in a common garden plantation at the Univ. Toronto’s observatory. I met Vic Timmer (UT forest soils) on a job interview, and saw the site – then found money to send Beth Anne (with Dave Valentine to help) up to sample the soils (as an undergrad research thesis).
The Colorado Years:
1. First project, with department-controlled funds, was to work with Skip Smith to put in some single-tree fertilization trials across an age sequence on the Medicine Bow. The U. Wyo. Folks had done lots of biogeochemical studies, but hadn’t included any test of whether nutrients mattered for productivity. Straightforward idea, included the graphic analysis approach I picked up from Vic Timmer in Toronto.
2. Eucalyptus and Albizia in Hawaii – Forest Science sent me a manuscript by Dean DeBell to review. He had replicated plantations (4 yr old) of Eucalyptus and Albizia, in monocultures and mixtures. He found that the mixtures had greater production than the monocultures – I knew from a silviculture class that this almost never happens; I suspected from my alder work that it could happen when a slow-growing N fixer was mixed with a fast-growing species on a poor soil; but the complete study was not available for alder/conifers. A few months later, I had a trip to the Philippines – and before I left I called Dean and asked if someone could give me a tour of the Hawaii sites. Dean made the contacts, I was enthralled by the plots (and the rapid tree growth around the area), and was invited to develop ideas to use the plots. I had no funds, other than Kris Dunkin’s (MS student) stipend; I donated some consulting income to CSU’s foundation to cover airfare for Chris and 2 assistants to sample the soil (she did 15N pool dilution stuff), and to set up litterfall traps. Dean arranged for the Forest Service folks to collect the litterfall and send it to us. Over the next few years, other students used the plots (Chuck Rhoades determined why pH differed between species; Xiaoming Zou evaluated differences in worm populations that he noticed while sampling the soils for a new method he developed to look at P transformations; Diana Garcia did a thorough assessment of N at these plots and some others I ran into during a sabbatical there; I did a bioassay of soil fertility and nutrient limitation; Christian Giardina did some P fractionation work to go along with some resin bag data from the field; we looked at mixed-species effects by crossing the boundary between pure species plots, and Mike Ryan and I cobbled together an NPP story; Kaye, Resh, Kaye, and Chimner did a final sampling of the mixed species trial (for soils and biomass). The finding of higher P cycling under Albizia, as well as N cycling, intrigued me, and I tried 3 times to get a project funded, but didn’t come close to success.
Meanwhile – Daves Wedin and Tilman were claiming that species effects on soil N mineralization were ephemeral – big at the start of an incubation, and small at the end. Neal Scott found the same thing in his PhD work. So I wanted to cut down the Euc/Albizia plots and start a new rotation – with the same and reciprocal species – to see how long the legacy of the first generation lasted (and how easily fertilization could overcome the legacy). The forestry interests on Hawaii had declined to near 0, I discussed some preliminary ideas with Bob Powers and Ariel Lugo (hoping to do a second rotation study in Hawaii and PR), but gave up because of the insurmountable logistic challenge. The idea was revived by Randy Senock, who proposed that he could handle the on-site work from UH-Hilo, and that NSF should give the proposal special consideration given UH-Hilo’s undergraduate minority college status. The first proposal went in and NSF “forgot” to consider UHH’s special status; Randy took over revising the second draft and made sure they didn’t forget – and we were funded (the special status of UHH tipped the balance for us).
3. Land-use change and soil C dynamics with afforestation. While sampling the Euc/Albizia soils for the bioassay with Felipe Garcia, an idea came up that the sugarcane would have labeled the soil with a 13C signature that could be followed into tree plantations. (Felipe had done some similar work – deforestation and conversion to C4 pastures in Mexico.) Michael Bashkin developed this for his MS thesis (looking at conversion of C4 cane fields to Euc. plantations). Sigrid Resh used the idea to examine the effects of Eucs. vs. N-fixer trees. Sigrid also used sites in Puerto Rico – when Xiaoming Zou took a job in PR, I went to visit – and asked to visit the Toa Baja site that I’d read about (Parrotta and colleagues had estimated N fixation by labeling the soil with 15N, using a variety of species). I also wanted to see the Lajas plots that Ariel Lugo had established (I read about them in a paper by Deane Wang). Years later, Sigrid was able to use the plots to test the N-fixer effects far beyond the andic Hawaiian soils.
3. Hawaii production ecology – The ESA had an annual meeting in Honolulu, and Jim Fownes presented some figures that claimed tropical plantations would reach a peak in growth at age 2 yr (or so) and then decline. I’d been interested in the causes of decline of growth (from the growth analysis of the Cascade Head alder/conifer plots in the early 80s, and testing the nutrient limitation idea with Skip Smith in Wyoming, and at Fraser). I thought the decline in growth resulted from the fertilization regime (fertilization stopped before the decline), but Jim thought it was a real phenomenon unrelated to nutrient supply (Jim's recollections on this differ somewhat from mine). We decided it was worth testing, and Mike Ryan was dragged in because of his interest in the question, and experience in assessing tree respiration and production. Mike funded a proposal-writing meeting, where we developed our alternate hypotheses, and ideas of how we’d like to test them. As the project rolled along, Barbara Bond and Mike came up with the hydraulic limitation idea, and this was incorporated into our renewal proposal. Holly Barnard's MS work showed the hydraulic ideas were not important to our growth decline. So, it looked like we disproved all possible explanations about how resource supply and use could lead to a stand-level decline.
Key lessons from the project (from Jim Fownes):
A. Multiple processes may interact to produce universal patterns (such as the decline in stem growth) -- universal patterns need not be driven by single universal processes.
B. There is great value (and fun) in arguing ideas with colleagues -- the ability to argue ideas and data without taking the arguments personally ranks high on the list of factors leading to success in science.
C. Keeping an open mind about experiments is valuable, even if one or another expert is convinced he knows the answer; always consider alternative hypotheses (including ones from other perspectives -- nutrient supply doesn't explain everything on the planet).
4. Brazil production ecology -- The Hawaii production experiment found no suitable explanations for stand-level growth decline, and that led me to develop an idea about single-tree components of stand-level growth decline: non-dominant trees continue to use resources after canopy closure, but not as efficiently as they did before canopy closure -- so stand level growth declines despite continued high use of resources.. After some testing in our Hawaii plots, we developed an idea to test in clonal plantations in Brazil (where forest industry makes forestry research much easier than in Hawaii). Clonal stands have a very small range in diameters, and therefore should not show these single-tree declines in resource use efficiency. I thought we could test this by testing for a lack of decline in irrigated and fertilized clones in Brazil, vs. the same clonal material in plots that have a range of stem sizes because of initial clipping of some of the seedlings at the time of planting. Christian Giardina suggested it would be easier to just stagger the time of planting, rather than clip seedlings. Through Jose Stape's contacts in Brazil, we were able to have 5 forest companies put in this experiment for us.
5. Alaska: Bob Stottlemyer had been working in Alaska, and thought that some terrestrial ecology would be a good complement to his water quality work. He invited me up to AK to scout around; a bad fire season gave us no helicopter time… but meanwhile, I left Bob with some resin bags that he took to another site later in the season and stuck in the ground (as well as buried soil cores) in a range of ecosystem types (willow, tundra, spruce, alder). The next year, Bob decided to fund a grad student and assistant; we all flew up to this site I had never seen, collected the resin bags and came up with ideas to keep the 2-person crew busy for the summer. Things we noticed in simply walking around: a terrace had it’s edge exposed by river erosion, and revealed a gradient in the thickness of the silt cap over gravel (so we designed a study to document species diversity, and N and water limitation as a function of silt depth); the boundaries of spruce/tundra had lots of small trees, suggesting invasion of the tundra by trees (so we designed a tree invasion study). We also tried out other ideas – single-tree fertilization to determine nutrient limitations, soil warming (and N mineralization) with plastic. In a later year, Mike Arbaugh wanted some tree cores for dendrochronology work, and in addition to this site he took advice to sample a site farther north. He reported back that the northern site had good successional terraces, and I checked it out a year later, and we developed some succession work as a result. I tried to sell a proposal to NSF, about a historic role of moose in determining the timing of spruce dominance, with a test of a litter quality idea from John Pastor… but it wasn’t funded (and that was just as well, as I had misinterpreted some of the terrace development stuff – I had the story wrong!)
6. RMNP, Green/Yampa Rivers – these projects have come about by other people having funds, thinking nutrients might be important, and inviting us to come measure things. The ideas are either straightforward (measure soil N inside and outside exclosures), come from the outside collaborators (cottonwood stories on the Green and Yampa), or through discussions.
7. Some CSU student projects not covered above:
Kris Dunkin – the idea development for the Hawaii Euc/Albizia study on soil N mineralization and litterfall nutrients was straightforward given the plots.
Xiaoming Zou – developed a variety of ideas, dealing with P across gradients in Costa Rica, P in agroforestry in China, and finally settled on developing and testing a new method for P transformations – tested it on as broad a range of soils as he could obtain. While digging soil pits in Hawaii, noticed some big differences in worms between plots, and developed a project in that direction.
Laura Stump – the USFS offered support for a student to do something about nutrients at the Fraser Experimental Forest. I had disliked Pastor’s use of lignin:N in litter to gauge N mineralization in forests – esp. since I thought a big part of the story was mineralization of humified mineral soil material, not forest floor. I developed the idea to look at lignin:N in fresh litter (leaves and roots) vs. in-field net N mineralization of forest floors and mineral soils; reading a paper from Univ. Alberta gave us the idea to do an accelerated decomposition study in the lab… Laura Stump did a great job, and ended up supporting Pastor’s idea (at least across species), much to my chagrin.
Chuck Rhoades – developed a plan to look at the role of resource use efficiency in agroforestry systems in India (looking at how efficiently trees vs. row crops used nutrients) – was almost ready to go and was then un-invited when a personnel shakeup occurred in India. As a backup project, he applied the soil titration approach that Valentine and I had developed to the Hawaii Euc/Albizia plots.
Neal Scott – The USFS offered support for a grad student to work on something about nutrients at Fraser; Neal did some initial work (using 15N to look at N transfer from mineral soil to forest floors), but never found a good topic there. He ended up developing ideas about the role of species in changing aggregation in soils, and the impacts this might have on N mineralization. He came up with some common garden sites for grasslands and forests; he ended up concluding that species affect aggregation, and species affect N mineralization (at least short-term), but aggregation is not the mechanism by which species affect mineralization.
Kuni Suzuki – I read a paper by Bill Romme on aspen in Yellowstone, and sent him a letter asking his opinion about some similar work I wanted to do in Rocky. Bill said great idea, and told me that Bill Baker had already done it. I asked Bill Baker, and he sent me a preprint of their work. I decided it would then be interesting to look outside the park, where hunting might give a different twist on the elk/aspen story. I had some funds to support Kuni Suzuki, and this seemed like a good project that wouldn’t cost too much (and Stohlgren ended up finding more support). Kuni designed a sampling scheme for the aspen stands, and a sampling protocol within the stands – modeled on the work of Romme and Baker. Kuni found very quickly that most aspen stands outside the Park regenerated very nicely, and we decided he should reduce his sample size in the National Forest, and sample inside the Park in areas that Baker had not sampled.