About Rosemary Hartman

Rosie is a just completed her PhD in Ecology studying the effect of introduced trout on amphibians in mountain lakes. Is now working for the California Department of Fish and Wildlife trying where she is planning monitoring of tidal restoration sites and writing a data management plan.

E-mail 


Data management – actually easier than herding cats

It’s a new year, new grant cycle, time to think about everyone’s favorite topic – Data management! OK, it’s probably not the most fun thing you will ever do. If you really love databases, Endnote libraries, data dictionaries, and quality assurance plans, you will have great job security later in life. However, if you are like me you started grad school with a vague proficiency in Excel and some idea that you should take good lab notes. Managing data in any organized fashion was never as fun or interesting as collecting it. I was sure to remember what I did, why I did it, and where I put it, right?

You can get away with sloppy data management with a small project, but you should start good habits early. Even that pilot study needs a little help from the data organization cat!

No. Not really. It is amazing how difficult it is to remember why I didn’t catch any mice at Big Marshy Lake, or what LITTER stood for when I came up with that acronym. If you want to produce useful data that allows you to write scientifically defensible publications and share your data with others, you need a data management plan.

Not convinced? Well, more to the point, many funding agencies now require a data management section in grant proposals where you describe your plans to efficiently collect, QC, analyze, store, and share your data.

So what is data management? I recently attended two sessions at ESA in Sacramento on tools and tips for managing ecological data. The organizers were from DataONE,  (Data Observation Network for Earth) which is an organization that searches data stored across many different data repositories and organizes multiple tools and resources for managing and analyzing data. They broke down the process of data management and introduced tools to make it easier. I have attempted to compile their work into guidelines for your own data/cat management plans.

The data management life cycle, from USGS's data management website.

The data management life cycle, from USGS’s data management website.

Step 1: Plan

Plan your data collection and management before you start. You will undoubtedly make changes as you go, but starting with a plan will prevent avoidable mistakes.

Developing your data management plan is hard work. It should be done in a very comfortable chair.

  • Start by looking at the whole data life cycle, you want to plan every step before you collect your first data point. Check out https://dmptool.org/dashboard  , it’s like TurboTax for data management plans. They even include specific requirements that funding agencies require in management plans for their grant proposals.
  • What protocols will you use? Your data should be comparable with others, so try to use established protocols and methods when you can instead of developing your own from scratch.
  • What format will you use to record the data? Will you use paper data sheets or direct computer entry?
  • Who will collect your data? Training interns/kittens before you start to make sure everyone is interpreting and recordings things the same way will help avoid errors. Training them in good data management skills is also critical  (I wish I had told my kittens to always write the date on their data sheets… I have no memory of when we collected those hairballs at the neighbors catnip patch).
  • Start metadata early! Make a “data dictionary” of common terms and abbreviations that you can refer to in the field.
  •  It is important to think how others might use the data, so include anything that might seem redundant or unnecessary to record for a single study, but puts the data in context with other studies of its type.
  • Where will the data be stored? Decide on a repository and whether you will have to put any restrictions on its use (see Preservation section below).

Step 2: Acquire

Along the way: Manage quality

(also known as Quality Assurance/Quality Control, or QA/QC)

This cat makes a fatal mistake in quality control. Never shred your mistakes. Fix them or omit them from analysis, but keep a record of the original version.

  • QA/QC begins in the planning stage and follows you throughout the data life cycle.
  • Do most of the work before data are collected:
    • Assign responsibility for quality control to one cat in the lab/field
    • Define and enforce data collection standards
    • Minimize repeat entries
    • Learn to use databases effectively (Most felines don’t really know how relational databases work). Using relational databases minimizes repetition and allows you to query the information you need for analysis in the correct format quickly and easily.
  • Use tools such as forms and data-checking techniques in most database programs
  • If you are managing a large group of kittens, perform data or lab audits to make sure all felines are collecting data the same way
  • Bring in an expert to check your work, or send a certain percentage of your samples to an alternate lab for confirmation.
  • Search data for outliers before analysis. Some techniques for finding outliers include basic graphic techniques such as histograms, scatter plots, and quantitative checks such as comparing median with the mean.
  • Deal with outliers on a case-by-case basis
  • Always document changes made to the data.

Along the way: Describe your data (Metadata!)

  •  Metadata is the who, what, where, when, why and how of your data. Like QA/QC, this should follow you throughout the data life cycle
  •  Record your planning process
  • Record what protocols were used to collect all your data, who collected it, when it was collected, and where it was collected
  • Record what quality control procedures were used on it, and any changes made after data was collected
  •  Record where the data will be stored and who will be allowed access to it.
  • A document with all of these metadata should be included with your data whenever you do anything!
  • This tool can help you put your metadata into a standardized format and stop you from forgetting anything: https://knb.ecoinformatics.org/#tools/morpho

Along the way: Back up and Secure

  • Your data is precious. Do not leave your entire dissertation sitting on a thumb drive that gets lost in a hole in your pocket. Do not trust it to your hard drive. Keep it in multiple locations, one of which should be an on-line service that is accessible anywhere.
  • Scan or make copies of paper data sheets
  • Back up your files in non-proprietary formats such as .csv and .txt. This will make it easier for others to use and will add to its longevity.
  • Store metadata with your data.

Step 3 and 4: Process and analyze your data

Don’t forget to record exactly when and where cats fell asleep on your laptop during the data analysis stage.

  •  Hopefully you planned what transformations and statistical analyses you are going to do in the planning part of the data life cycle.  Now is the time to get cracking!
  • Do not neglect your metadata during this stage in the process. Record what analyses you tried that did not work out as well as the ones you plan on using so that you do not repeat your mistakes.
  • I won’t go further into how to do data analysis here, that’s a whole ‘nother bag of cats…

Step 5: Preserve

  • If you have been backing up your data regularly, you should be half-way there. However, back-ups are designed to restore what you are working on in case you lose it, while archives and repositories are built for long-term storage and reuse by others.

Choose your data storage wisely.

  • Re-evaluate your documentation. Would an outside researcher be able to recreate what you did? Is there sufficient information to place your data in context?
  • Store your data in an on-line repository and include a data attribution file with full information on who produced the data set and who should be contacted for more information.  These repositories also provide a  DOI (Digital Object Identifier), to make your data easier to cite and discover.  Repositories include:
  • Consider licensing and legal issues. For example, many federally funded projects require data to be publicly available. Some open-access journals (PLOS and others) also require data to be publicly available. However, data concerning human subjects or location data on species of special concern may be sensitive. (more on legal and privacy issues from DataOne’s policy guide)

DOIs allow other people who use your data to cite it properly.

Step 6: Share your data

  • Just because your data is up in a publicly available repository doesn’t mean people will be able to find it or use it.
  • Include information on where your data can be found in any publications you produce using it.
  • Submit your dataset to Data Portals and Catalogs to them more visible and more likely to be employed by others. Data Catalogs and Portals (Like DataOne) provide searchable directories of data and usually include many data repositories.
  • Tweet it! Facebook it! Social media is the way of the future!

 

Lots of scientists use twitter these days to publicize new publications, including data sets.

 

More resources on data management:

USGS data management website:

http://www.usgs.gov/datamanagement/index.php

Cool paper on data collection:

http://www.esajournals.org/doi/pdf/10.1890/0012-9623-90.2.205

Data Management Plan Tool:

https://dmptool.org/dashboard

Data management best practices:

https://www.dataone.org/sites/all/documents/DataONE_BP_Primer_020212.pdf

How to get an ecology PhD in four years (a 12-step program)

Step 1: Find a great program.  One with a supportive, fun student scene and an open academic culture. UC Davis, of course, is da bomb on both these fronts, though I’m sure there are other good programs out there. Avoid places where students compete with each other, feel uncomfortable talking to faculty members, or are insecure about wearing pink.

The UC Davis Graduate Group in Ecology.

Step 2: Find a great adviser. Wiser men/women than I have already said that your relationship with your major professor is the thing that will make or break your graduate school experience. It is easy to fall into the trap of wanting to be the student of the big name in the field, but if they do not have time to help you, or they make you cry, having access to their notoriety is not worth it. Because you have already completed Step 1, you are part of a wonderfully open academic culture where you can talk to any professor you like (but see steps 4 and 10 for hints on how this works). Therefore, find a major professor who will listen to you, support you, and give constructive criticism on your work. Form temporary collaborations with the big names.

My adviser is the one in the bomber hat.

My adviser is the one in the bomber hat.

Step 3: Steal other people’s ideas. Unless you have been working in your study system for years, you are a genius, or both (I was neither), chances are the research ideas you come up with during your first month at grad school in order to apply for the NSF GFRP won’t work out. You will probably spend your first year or two starting from scratch two or three times. That’s fine! It’s part of the learning process. However, if you want to get out in four years instead of seven, then you may want to start with an idea that has been partially worked out. Ask your professors, senior grad students, former grad students “What are the big questions that need to be worked out? What do you think needs to be done but you don’t have time to work on?”. Once you get started you will discover your own questions that are useful and feasible, but starting with someone else’s ideas can really help you get going.

Step 4: Be really annoying. Pester your professors constantly. Not just those on your committee or those teaching your classes, but EVERYONE who might have something you want. Talk to the seminar speakers, track down emeritus professors, send an e-mail to the researcher who wrote that really cool paper from New Zealand, and don’t take no for an answer. While you may feel like they have better things to do than help you out, remember that most people love talking about themselves and what they research. Academics live for flattery; so coming to them for their wisdom is paying them a compliment.

Step 5: Write a lot of grants. Coming in to a lab with good funding means you can theoretically spend your time working on research instead of begging for money, but the process of grant writing can be very valuable for framing your ideas, thoroughly reviewing the literature, and making valuable connections. I funded my own research partially through scholarships given out by various fly-fishing clubs, and getting the chance to present my research to them was one of the highlights of my time here.  Plus, if you’re lucky, grant writing gets you some money too!

Step 6: Find lots of great collaborators. This includes undergraduates, who will not only sit for hours sorting bugs for you, carry your gear up mountains, and come in on weekends to watch tadpoles with you, but also frequently have great ideas. This also includes other graduate students. You probably are spending happy hour griping about your research over a beer with them anyway, so going in together on a project, or at least trading field help with them, just makes good sense. More difficult but also more rewarding, is looking for collaborators outside the university. Scientists who work for state or federal agencies have a huge wealth of practical knowledge about management of local ecosystems. This is invaluable to a graduate student. (They also often have an inside track to getting scientific collecting permits and using your research in policy).

DSCF0107

Some of my favorite colaborators.

 

Step 7: Go outside your comfort zone. If you are a field ecologist, it is easy to want to construct a dissertation comprised of three related field studies. However, some of the questions you want to answer may be better suited to a lab experiment or a computer model. If you approach your question from multiple angles rather than trying to takle multiple questions, the final product will be more well-rounded and rewarding. Furthermore, asking a similar question with multiple techniques can take less time than developing several questions best answered by a single technique. My dissertation involved an ecological field study, a sociological survey, a laboratory experiment, and a computer model. This not only forced me to learn multiple types of scientific techniques, more importantly it taught me how to teach myself scientific techniques quickly and effectively. It also taught me how to find collaborators (See steps 4, 6, and 10)!

Step 8: Trust your own ideas (your professors don’t know everything). While you should listen to everything the tenured Distinguished Professor in Sumthinorother says, you do not have to do what they say. It is your dissertation and if you do not think you can take care of 500 baby frogs at once, do not let them tell you that you can. Similarly, if they say no one cares about fresh water sponges and YOU care about fresh water sponges, study fresh water sponges. You are much more likely to succeed. There is also a good chance you will prove them wrong and the world will see how important fresh water sponges are.

Did you know there was such a thing as fresh water sponges? They have symbiotic algae too!

 

Step 9: Be insane. Seriously, if you want to enjoy your grad school experience with a stress-free lifestyle, I recommend ignoring all of these steps and prioritizing vacations, $1 pint night, Battlestar Gallactica, Burning Man, and sleeping in. You should be out in less than eight years with an intact psyche. Don’t try and get a PhD in ecology in four years.

Step 10: Be really annoying some more. No one answers their e-mails. It is a fact of life. Keep badgering them. Write e-mails with “Are you getting my e-mails?” in the subject line. That usually gets you something. Stand up for yourself. If you are on top of things, do not wait months for your last committee member to get back to you with comments on dissertation chapter 3. Just submit it to a journal. Once it’s accepted by the peer review process, your committee pretty much has to accept it.

Step 11: Have an outlet. While this may appear to contradict Step 9, mental breakdowns can interrupt field seasons, cause mistakes in lab, delay writing, and cause your collaborators to doubt your abilities. Therefore, take a weekend or two in the mountains. Take a pottery class at the craft center. Go to costume dance parties. Frequently. Just make sure to get back before midnight or you turn into a pumpkin.

frog2

Somehow my study system kept creeping into my art.

Step 12: Find a job. At the end of my PhD I applied for a bevy of government agency and non-profit jobs because I wanted to have the most constructive impact on conservation in California that I could, and I landed a job with the California Department of Fish and Wildlife.  You may want to try the academic route and apply for post-docs and professorships, but the application process is often longer, harder, and more competitive. Since I am just beginning the “what comes next phase” I can offer less advice on step 12 than the steps leading up to the four-year PhD, but when in doubt, follow your dreams. It’s cliché, but if you can get a PhD in four years you can probably tackle just about anything.

Care and feeding of the Cascades frog

When I started this summer’s science experiment, I foolishly though that my hypothesis, methods, and data analysis would be the most important decisions I had to make. Little did I know I was not starting a science experiment, but adopting a pet.

Instead of going to the local animal shelter and picking out a kitten, I had to prepare for not one, not two, not twenty, but 600 new pets that would require my care and attention. To my dismay, there was not a single book in the entire University of California Library system on care and feeding of Cascades frogs. The closest the on-line forums came was bullfrogs, and they recommended feeding the tadpoles “boiled lettuce”. I ran around to my friends and fellow researchers, but the only group who had raised Cascades frogs in captivity before had horror stories of harmful algal blooms, massive die-offs, and tiny froglets with no bone mass.

This was beginning to look more difficult than I made it sound in my grant proposal. But they gave me the grant, and the frogs were laying eggs up in the mountains, so if I wanted an experiment I was going to have to figure it out.

After 25 miles of hiking in the snow (up hill both ways), I had my eggs, and now it was time to hatch them. Setting up my tanks involved an interesting shopping trip:

o      3 tubes silicone adhesive

o      2 rolls black plastic mesh

o      3 aquarium nets

o      4 paint buckets

o      12 wash basins

o      48 stems plastic aquarium plants

o      150 feet airline hose

o      24 air stones

o      1 bag organic meadow hay

o      5 lbs organic rabbit chow

o      3 lbs frozen spinach

o      8 oz spirulina flakes

o      1 lb dark chocolate (I’m going to need it)

Interestingly enough, all of the above was available at Davis Ace Hardware.

Feeding time at the tadpole tank.

Feeding time at the tadpole tank.

It took me a few days to figure out the right rate of water flow and system of standpipes to keep the oxygen in my tanks high but avoid clogging the drains with rabbit chow. To make matters more interesting, the new baby tadpoles hatched and decided that they actually were small enough to go through the mesh I had covering the drains. Another trip to Ace and smaller mesh finally got me a system that seemed to work. But their insistence on being too small for the mesh should have warned me that these tadpoles were the most ornery creatures on the face of the planet.

For one thing, they grew much too fast. Every time I turned around they had doubled in size and soon they were threatening to crawl out of the tanks. This was irksome because I was supposed to conduct 84 behavioral assays (each of which take 30 minutes) while they were all 2cm in length. Well, 2-4 cm is close enough, right?

I had planned my summer around the fact that they were supposed to start metamorphosing around three months of age. Thus, I planned to leave for a conference on the east coast when they were two months old. I could have my undergrads check on them every day and toss a little more spinach in their tanks during the eight days I would be gone. No problem!

Then when they were one month old they started growing legs. But it takes legs a while to grow, right? But they kept getting bigger and bigger and their legs were getting longer and longer. At six weeks their arms started popping out. Seven weeks (and three days before I was supposed to leave), their tails started shrinking. I was about to go from several hundred tadpoles who were happy to sit in their tanks and eat spinach to several hundred frogs who would be hopping all over the room, probably falling in the sturgeon tanks next door, and wreaking carnage on the cricket experiments down the hall.

Not really a tadpole anymore...

Not really a tadpole anymore…

Many frantic phone calls later, I arranged an animal room (trying not to think about the MRSA infected rabbits that were there before). I got coolers ready, washed approximately one grillion plastic shoeboxes to house frogs, cut up plastic weighboats for frog houses, hauled bottles of flightless fruit flies across campus (yummy), and wrote a lot of (hopefully) very clear instructions. Thanks to my hard-working undergraduate assistants and magnanimous adviser, I was able to go to the conference, but I’m not sure I got much out of it because I was constantly checking my e-mail, waiting desperately for the daily frog report telling me who had metamorphosed. I suddenly had great sympathy for any father who has been told to wait outside while his wife gave birth. Except instead of one baby, I ended up with several hundred.

And labor lasted three weeks!

Everyone said metamorphosis would be the worst of it. I could see their point. We  had to drive several hundred new froglets from the tadpole tanks across campus to the frog room, weigh them, measure them, and get them settled in to their new home. It also meant we had to take care of feeding the tadpoles their daily spinach and checking water quality as well as taking care of the frogs and ferrying the new guys back and forth. But no, these frogs were too ornery to let that be the worst of it.

The Amphibian ICU

The Amphibian ICU

Because once they metamorphosed, half of them got sick. Very sick. Their bellies and legs were red with inflammation, they were losing skin in great sheets, and they stopped eating. I dreaded coming in because I knew someone would have died. The worst was seeing frogs is pain and realizing I had to put them out of their misery. We did this with an overdose of anesthetic, so they certainly weren’t in pain, but I was in agony with insecurities over whether I had made the right decision. Were they really in that much pain? Would they have recovered? I knew then that I could never be a vet. It was just too many small lives who couldn’t tell me where it hurt.

Suffice it to say, we immediately called in the real vets. One of the most heart-warming things I have ever seen was four grown women (two vets, my advisor, and myself) all huddling around a 1.5g frog trying to figure out how to give it a shot. “Inject 0.05 mL per gram frog into the subcutaneous space” sounds easy enough when it is written on a bottle of antibiotics, but when the frog is the size of your thumbnail, it is not so easy. This was not “take two aspirin and call me in the morning” medicine. This was staying up all night searching the relevant literature for ideas, coming in multiple times a day to check on the frogs, and hoping by shear force of will that we could cure these frogs. If will were enough, those frogs would have been fine. I owe a lot to the vets who didn’t even know these frogs who were still helping day and night to get them well again.

In the end, we never found out what was wrong with them. But gradually they recovered. We lost many good friends that week, but we continue our experiment in their honor.

All good science starts with a question. You then develop a hypothesis, an answer to your question. After careful research you decide on a method to test your hypothesis. In this case, the method involved live, recently-metamorphosed Cascades frogs. Even the best hypothesis, and the most rigorously designed experiment in biology hinges on the wonderful, fantastic enigma that is a living thing. Biology is the study of life, and life is so complicated that testing just one question is impossible. To test your question you have to master the most basic part of biology – how to keep something alive, and whether or not you answer your question you will learn the most valuable part of science – how to love what you study.

Ain't he cute?
Ain’t he cute?