Data Challenge: Is It Mature Yet?


#1

Corn has clearly defined growth stages.

Vegetative growth is defined by how many leaves have emerged. When the ‘black layer’ forms on the seed, the maximum amount of starch is in the seed and harvest is merely a matter of waiting for the moisture content to decrease (standard weight is based on 15.5% moisture). Growth rate is based on ‘growing degree days’. If you know the temperature, you know how fast the corn will grow.

Why is this significant, as most of us are not going to grow corn in the Food Computer (it won’t fit)? The reason is phenotypic data. With corn, there are clearly defined stages of growth, and standard measurements of grain yield (but was that metric or English tons?). I do not see similar precise definitions for lettuce, and most other vegetable crops. Without such standards, comparing measurements between experiments is almost meaningless.
I am sitting beside a nice head of lettuce right now, but can I say it is ‘done’? I could have picked it several days ago, or I could let it sit for a few more days until I am in the mood for a salad. Short of tearing it apart and measuring the number of leaves (how about the ones that dried up?) and the total surface area of the leaves, how do I compare my lettuce to yours? Do we always weight it on the 31st day (but are we growing at the same temperature?)? To not have clearly defined genomic traits is like measuring temperature as ‘cold’ and ‘warm’; it is a measurement, but it is neither precise nor accurate. As we know from mathematics, when you combine numbers, the number with the least significant digits determines the precision of the outcome. Right now we have precise sensors for the Food Computer (though I find two thermometers can be off from each other by almost a degree), but we lack precise definitions for plants. It is not sufficient to say that a mature head of lettuce is ‘the size of two fists’. How ‘tight’ is the head? How ‘stiff/crisp’ are the leaves?
There is a lot to the engineering of a Food Computer and writing code for a Raspberry Pi, but these are problems with known solutions; by comparison, the ontology of phenotypic traits (at least for lettuce) is a relatively unknown problem with no clear solution at this time. These are the problems that literally wake me up in the middle of the night. Without some answers (that we can all agree upon), the Food Computer will be reduced to a very expensive toy.
What do we mean by a ‘mature head of lettuce’?


Getting Recipe Data - the Old Fashioned Way
Sensor Data Modeling
Beyon Here Be Dragons: Ontology and Semantics
#2

Beyond a single harvest event like a store bought head of lettuce or 3 stalks of corn, you can choose to harvest over time. You could choose to continuously harvest the lettuce by regularly harvesting some percent of leaves at some interval. Say 70% of the leaves every 10 days. Depending on a wide range of factors you might get 8 leaves over that 10 days or maybe you get 12.

There are also possibilities for tuning that regular harvest. Maybe you harvest 60% and allow for a higher percentage of leaves to continue the growth. Or maybe its best to harvest continually harvest every 7 days.

On top of that, all the variables that go into growing those additional leaves might be slightly different for a plant that is trying to recover from loosing leaves vs a plant that is growing an initial head.

There is an almost incalculable number of possibilities with just a single strain of lettuce. And different strains could have different behaviors under a multiple harvest scenario.

And growth is just one aspect, what about nutritional content or flavor.

I have to believe that part of this project is building up a minimum viable set of laboratories. That would be everyone who has taken on the challenge of building a food computer. I think it will take several years of build up just to get to the point were you might be able to have statistics of 1000 food computer trials that all grew the same strain of lettuce with slightly different growing conditions.


#3

Funny you should say that, I just ripped the bottom leaves off one of my plants last night for dinner, and now trying to figure out if I should let the top keep growing, or encourage the shoot coming up from lower down.


This is the point I keep trying to make, if you don’t have clearly defined goals, then you can head off into any direction, and data collection goes crazy (and is virtually worthless). I am seeing this as possibly a two part question:

  1. What are the key research questions that need to be initially answered? I am thinking we first need to define what it takes to repeatably produce a ‘good crop’ (undefined) of lettuce and other key plants.

  2. Are there any projects that could be taken on with something less than the full blown food computer? What is the minimal set-up needed to give valuable data?
    I can/am growing lettuce with a coffee jar, light and timer, and an air-stone. After a couple of temperature and humidity readings, I know these are stable and do not need to continuously monitor these. This is an open-air environment so temperature and humidity cannot be varied. I could move this to a bus tub and expand this growing set-up. I do not think this is enough for research, as I think that requires a controlled environment where temperature and humidity can be controlled (and varied). I can envision a three step growth path to the Food Computer:

  3. Coffee jar or bus tub, air pump, timer and light. This could be enhanced with an Arduino to occasionally take some sensor readings (temp/humidity, LUX). The Arduino could run basic sketches, with the results recorded on paper or spreadsheet. I think one or two GE Bright Stiks (100w equiv) in a reflector would do for lighting, or possibly jump up to a CREE COB. This would give some basic data, likely not research grade, but useful for casual growers.

  4. Next level would be to make an enclosure (Foam Farm?). This would require the addition of a RaspberryPi and fans (circulation and ventilation). Software could be the current brain, but I think it might be better to make some modifications (UI with user data input). This is potentially serious data, though most of the significant data would be manually input at this time (phenotype stuff).

  5. Final level would probably be the Food Computer - better enclosure and adding pH and dosing controls.

Again, it is back to defining goals. Does this sound like something worth pursuing?


Growing food: I just ordered a MicroGrow Kit from Hamama
#4

That really depends on who’s asking :slight_smile:
For a researcher answer to this question will be different than for a DIY gardener, and a commercial grower will have yet another answer. If the FC is a platform all of these entities will be able to use and share the data in a meaningful (to them) way.

Let’s work on an answer to your question from each perspective.

Is it mature yet?

As a home gardener:
In this situation, the FC does not need to answer the question in a definite, deterministic way. The user will either have to determine it himself, or the answer will be based on the time since the lettuce was planted.

As a commercial grower:
Most commercial growers will use a well-tested recipe, and they will make the determination using one or more properties (color, height, time, weight, volume) of the plant.

As a researcher:
In this case, the answer will really depend on what is it that the researcher is trying to determine. Are they comparing the growth rate of a specific plant in a different environment? Effect of light on the quality of the plant?

Now, how all of this affects data sharing?
One thing that will be necessary is to create different types of recipes.
For example, you could have the following categories

  • Food
  • Research
  • Seeds
  • Flowering

Each recipe would have maturity condition specified and the condition will depend on recipe type.
This would allow anyone to compare results only from recipes that relevant to what the grower is trying to achieve.


#5

@mariusz I think we will generally get to the same place, but for the sake of precision, I am going to have some disagreement (and possibly change my own definitions).
I want to separate procedures (what is done) from the motivation for doing something. One is objective (out there) and the other is subjective (the intent in my head).
A recipe should describe an objective agronomic procedure (temperature, humidity, light, fertilizer application), and may identify a result or goal (1 pound head of lettuce, maximum growth in 28 days, most cost effective way to grow a 1 pound head of lettuce). If I have a recipe for chocolate chip cookies, the recipe does not change if I am making them for a personal snack, a birthday party or selling them in a commercial bakery. There will however be different recipes if they are to be gluten free, chewy or extra crisp. In the same way that a ‘leaf’ recipe would be different from a ‘seed’ recipe for the same type of lettuce. However, I don’t want to have to re-label my recipe, if at the end of a 28 day research trial I pour balsamic vinegar and olive oil over my lettuce and have it for lunch.
The difference between food and research is not necessarily the recipe (agronomic protocol), but the observation and analysis protocols. This is where I see the need for a ‘plan’ that includes these three components (agronomic, observation and analytic). This is also where the intent or purpose may be stated; a research plan is different from a commercial plan (different purposes), even though they may share the exact same recipe.
I want to keep recipes defined as a discrete unit, separate from observations and analysis, because part of the OpenAg goal is to have research generate and share recipes that help non-research people grow food.

While ‘recipe’ needs to be clarified, it still doesn’t help to measurably and objectively define a mature head of lettuce (how many leaves, head weight, size, …?). The problem may be with the word ‘mature’, it may be a subjective and relative term in regards to salad greens production (and need to be dropped as a technical term); but that still leaves us with the problem of objective definitions of vegetative growth stages for lettuce. A lot of core plant physiology researchers have given up on using commercial plants, and instead use model organisms (Arabidopsis thaliana), but they are not that great with balsamic.


#6

@webbhm THANK YOU for mentioning Growing Degree Days!

Yes, this needs to be discussed in more detail. This thread was just mentioned / linked in the other thread. I don’t particularly care which thread (or even both) we continue talking about GDD but i think it is VERY important that we do so. The standard of “days-to-maturity” like in standard seed catalogs and in the old standard gardening paradigm is very inaccurate and misleading. For example a watermelon listed as 90 “days-to-maturity” grown in humid hot southern georgia when grown here in my high altitude shorter season with less heat during the summer growing season might take as long as 135 “days-to-maturity”. I only have something like 120 max. This is just an example of how region specific “days-to-maturity” is. I vote that we as a community and as a new generation of gardeners, plant breeders, and tinkerer farmers that we help to abolish and phase out this old outdated terminology and replace it with new terms. GDD days or heat unit requirements are needed for more accurate growing time estimates. I don’t know how GDD can be translated back into local “days-to-maturity” but i would like to learn how.

Another term that we should think about changing / abolishing is the term “Open-Pollinated Variety”. My understanding of the literal linguistic definition of that term is that it probably was origianlly meant to convey a plant being allowed to pollinate openly with the bees or “bee pollinated” ie. naturally outcrossing or naturally hybridizing. The modern usage of this term “Open-Pollinated” actually has been corrupted to mean the exact opposite or “highly-inbred” or “artificially kept pure” or “isolated from insect pollination”.

My friend and plant breeding collaborator Joseph Lofthouse has started using a new term for his widely famous grex and landrace bred crop varieties. He has started to use the term “Promiscuously Pollinated”. I like this term and hope it’s usage will catch on. You can learn more about Joseph if you want from his website. http://garden.lofthouse.com/


#7

My understanding of ‘open pollinated’ associates the term more with the genetic diversity of the plant population than with the method of pollination (though they are linked).
Prior to about 1930, all corn was ‘open pollinated’. A farmer grew a field of corn, and selected the best looking ears to hold back as seed for the next year. Each ear (and each kernel) was genetically different from every other one (like the genetics of a human population). Back then, if you looked at a field of corn you would see that each plant was slightly different in height - it did not have the uniformity you see today.
In the 1930’s hybrid seed corn became popular. It was discovered that seed produced by crossing two different varieties of corn out performed seed produced by breeding a variety to itself. Corn yield (bushels per acre) literally doubled over night. Before hybrid corn a farmer might get 25 bu/acre, with hybrids they got 50 bu/acre (farmers today average nearly 200 bu/acre) Corn breeding became a science, and all the seeds in a bag became genetically alike. The farmer’s field of corn was still wind pollinated, but the farmer no longer kept back seed for the next year. However, the seed breeder would control pollination (remove the tassels from one variety) to assure ‘closed pollination’ (though the plants were still pollinated by the wind).
Modern corn varieties literally start with a single seed which is inbred for genetic consistency till there is enough seed to be marketable. Consistency (especially consistency of health and yield) is the goal of the farmer, and preserving genetic diversity is left up to the universities and plant breeders who are concerned about maintaining a diversity of genes that may have potential use in the future.


#8

Yes, but you have just made my point exactly. “open pollinated” is confusing and does not reflect modern usage of what the term is trying to convey. “Open Pollinated” if being use to convey information about genetic diversity should be changed to something like “highly inbred” or “highly uniform” or something else. Not a term that implies about pollination techniques whether that implied meaning is intended or not. In this case it is not intended but goes along with it whether people like it or not.

And while you might be right about farmers wanting consistency in the past, i believe this is no longer true and will become less and less true as time goes on. We are currently in the midst of a food and agricultural change in society. We are no longer interested in uniform varieties except in some cases where that works better. Genetic diversity is what more and more people want in thier food. They want more color, nutrition, more variety. Few people care about uniformity anymore. The new age is about variety and choices. Grex breeding and Mass Crossing and Artificially created Landraces are the way to go for outdoor growing.

For a PFC in space uniformity is more important and indoors, but not necessarily required. In fact in a closed system where pests and diseases have no room to attach another plant species or variety high genetic diversity and low uniformity might actually be better in stemming off diseases and pests that are able to wipe out a whole production of food/plants like can be the case in “monoculture farming”. Monocultures are very susceptible to pests and diseases as the world around is never static. Protecting against biotic and abiotic stesses in PFC should be of high priority i would think. But i havn’t been active enough to know if it has been discussed here already or in great detail.


#9

You are into the old tension between quantity and quality.
Modern agriculture (based on number of acres grown) is focused around corn and soybeans (which is not human food!); veggies are about 10% by comparison. Farming is manufacturing, where consistency and repeatability increase efficiency and drive down cost. Machine harvesting of a plant that is a consistent height is more efficient than hand harvesting. The most recent manifestations of this are in grapes and olives (https://californiaoliveranch.com/olive-oil-101/how-its-made/).
Farmer’s Markets and the like go for the quality market, where the desire for variety and flavor justify the premium cost.
OpenAg is playing both sides. Hydroponics and growth chambers cost more than traditional gardening so need to play to the quality side, but we are also trying to maximize efficiency in that space. There are two markets here, those that are looking for home-based or small scale growing, and those looking toward large scale indoor, vertical farming. We are always playing off quantity and quality.