Every year, the mechanical engineering students at Purdue University—where I managed a research greenhouse for 20 years—compete in the national Rube Goldberg machine contest. Much like the board game Mousetrap many of us played as kids, Rube Goldberg teams build whimsical contraptions that accomplish a simple task with as many elaborate steps as possible. The 2012 Purdue team made a Guinness World Record by implementing more than 300 steps to pop a balloon. Every step must work perfectly for the proper outcome. None of the machines work every time. But they are so satisfying to watch when they do.
As plant cultivators, my colleagues and I often have been guilty of growing plants this way. Unlike the mechanical realm, so much of what takes place during plant cultivation is invisible. We can’t see the steps proceeding like clockwork: nutrient uptake, cell wall formation, floral initiation. Perhaps because of this, every time we learn a methodology, we add our own steps or ingredients based on our past experiences or some rare new malady we just heard about. The next person does the same. We don’t take the time to test for these necessities; we just want to not worry over them. The problem is that these plant-growth systems—the place, the parts, the protocol—are not robust. Everything must go so perfectly that the system can’t be adapted to change in conditions or in scale. When something goes wrong, we can’t easily identify the cause. It can’t be repeated at multiple facilities with the same outcome. It has become—in essence—a contraption.
How do we create plant-growth systems that can adapt to ever-changing plant strains, personnel and environment? (Even an indoor operation is susceptible to seasonal temperature and humidity flux.) How do we create methodology that can be explained and understood, so that a team of growers in one or more locations can be calibrated to that company’s lead grower? How can we collect data that will be meaningful today and three years from now as we track improvement? How can we make the team members understand the system so well that they work in concert with the plants and technology and can catch something afoul (what I believe to be the secret to robustness)?
Where to Start
To begin, robustness means your methods have been developed by controlled studies: when you trial a new miracle fertilizer ingredient, test multiple plants, capture quantifiable data rather than subjective eyeballing, and always have a set of plants that did not receive the miracle ingredient (a control group). If you’re spraying an ingredient dissolved in reverse-osmosis (RO) purified water on the leaves, you spray the control group plants with RO water without the ingredient to separate out its effect. You create robustness when your methods and development studies are then scaled up to replicate production level.
Here’s one of my own mistakes: I developed a research rice-growing protocol that used a smaller pot than is typically used and a root substrate that had a high cation (positively charged ion) exchange capacity, but the granules were sand-sized. The experimental plants were much greener and vigorous than the control group plants growing under standard rice production techniques. Six months later, however, I was tired of the extra steps required to keep the substrate from leaking out of the drainage holes and from up-righting top-heavy pots every morning because the small pots didn’t provide enough ballast! I wasn’t so clever after all.
The Role of Technology
Robust technology means utilizing components that are as simple as they need be, replaceable locally or stockpiled on site. Trust that I am a proponent of experimentation, and believe sensors, integrated equipment and environmental control software designed for growing plants are necessities for quality crops. However, they make us particularly vulnerable for failure if we don’t have a spare temperature sensor or pH probe that can be replaced on a Friday evening when they notoriously malfunction. High-tech without common sense is not robust.
We all know automation can be implemented to reduce labor and alert us to trouble after hours. Irrigation triggering and duration is best left to machines responding to our programming, and plant quality protected with remote alarming that calls us when fertilizer pH is out of bounds, temperature is too hot or power has been lost. Yet some of my favorite examples go the other direction, when humans augment technology to create a more foolproof growth system.
Most double-checking procedures are in place because these cultivation lessons have been learned the hard way. A smart manager sticks around to make sure that fan belt hasn’t slipped and the fan is really spinning, that those vents or HVAC dampers are really opening when called for. She occasionally comes in during the middle of the night to ensure all those light fixtures really shut off, or that a spigot isn’t leaking, causing an increase in humidity under the photoperiod cloth. To give her team visual cues if the irrigation system is working, she places one dripper in a transparent cup or graduated beaker so they can see how much was delivered the last time it initiated. They know the fertilizer injector is working because the water in the beaker is blue from the dye they added to their concentrate. The fertilizer concentrate has not run out because it is alarmed using an inline probe, but the alarm doesn’t go off because they left one vertical strip on the barrels unpainted so they could see the solution level from a distance. (This enables them to see the levels getting low before the alarm goes off.) The fertilizer was mixed properly because the manager has trained her employees to leave the bag labels on her desk when complete so she can check the quantity and formulation without standing over them as they do it. They know the nutrient film technique (NFT) pumps are running because of a simple indicator light illuminated on the pump’s exterior. They know which plants need more fertilizer because of weekly pour-through analysis of their root substrates, and they compare that data to past successful crops during this same season in years prior. Making Your Team Robust, Too In university research, every single plant—no matter how weak or spindly—represents data that the researcher requires for statistical analysis to test his hypothesis. Some research specimens are genetically unique and literally irreplaceable. How did we create robust plant-growth systems under such criteria? By rewarding good eyes. Acknowledge when someone on your team catches a problem that saves a plant or a crop, not with an award, but with simple, sincere words like, “Good catch with that clogged table drain,” spoken in front of their teammates. You’ll find that you get more of the behavior you reward in this manner. Learn a trick from commercial horticulture and assign a section of your facility to each grower. This practice promotes ownership and cross-training. When there is a lot to remember, provide written instructions; some people learn better this way than by hearing tasks explained, and those instructions are the basis of future standard operating procedures (SOPs).
Communicate the environmental settings to the team by placing graphs or setting lists in each grow room. Your team will learn when the temperature is normal and alert you when it is beginning to drift, or when the lights aren’t on when they are supposed to be. Use white marker boards for messages. Use color coding with signal tape or flags to clearly indicate plants that are not to be watered, or are to be discarded, or on which spider mites have been found and should not be handled.
Lastly, take the time to tell them stories, legends of past crops failed by preventable errors. How a plant strain or technology was discovered. What the best crop you ever saw looked like, and who grew it. How one person made a difference. Give them a vision of perfection, and they will view technology as the tool to achieve that vision, rather than just a machine they serve. Let them know that they are what separates a robust system from a contraption.
