I largely work with selfers. I like to work with them because I've always worked with fairly small populations in the hundreds rather than the thousands of plants. Most of my successes have been with self-pollinating plants. I did not really appreciate at all what John just tried to teach everyone here so I would find myself constantly treating selfers as if they were crossers and crossers as if they were selfers. Luckily, if you do that you can get somewhere with selfers, but with crossers you probably won't get anywhere. I'm going to talk about self-pollinating plants as well as some plants that are in the grey zone on the selfing side. John's going to talk about the complexities of working with crossers and go towards the grey zone on the crosser side. This grey zone is species like peppers about which one book will say less than 5% crossing occurs and another will say 25% crossing possibility. I bet that in my garden I get easily 50% crossing in peppers because I've got so many insects. The more insects, the more crossing, even in a selfer. The selfers have two evolutionary strategies­one is the conservative strategy and the other is the experimental strategy. We see this same thing in human populations and it must happen with ants, too. The conservatives, the selfers, when they hit on a good combination their reproductive biology tends to conserve that useful combination of genes and traits and through the selfing process maintain those qualities. Crossers are always experimenting, crossing with other genotypes, really trying to cover up their bad traits. They never get anything perfect, the poor crossers. They get it right and then the next generation they've gone back toward average. But a very successful self-pollinated population or variety tends to hang on to those traits until a really strong challenge comes along which forces it into experimentation.

I like to work with lettuce a lot and tomorrow I'm going to do a case study of my specific strategies in working with lettuce, including some OFRF-funded research that I did last year and hope to be doing for the next three years on investigating disease resistance and breeding new varieties of lettuce from things I find that have good disease resistance.

Self-pollinated plants, say lettuce heirlooms, tend to be pretty uniform but not perfectly uniform because they have been handed down informally from gardener to gardener or farmer to farmer and in those heirloom populations there will be a certain amount of genetic variation. All of the lettuces won't be exactly the same shape or some of them may have extra crinkling of the margins of the leaves, or variations in tipburn resistance. For example I got the heirloom Speckled Amish Butterhead and in the first growout in my climate 50% of the plants had tipburn, wow what a bummer! So, naturally, you don't let those make seed, you select for the traits that you want that you can see. That's all right but if the conditions are not those that will bring out tipburn, your selection will be ineffective. If the environmental conditions were such, and you had really high calcium levels in your soil and you were doing all the things as a farmer that would help prevent tipburn and nature was cooperating with you, then when you first grow out that population of lettuce, maybe you would see only 10% tipburn. And so I'd take those out, and save the seed and then the next time I grow it out maybe conditions are a lot worse or there's less calcium in the soil and then you see 40% tipburn. That's more than there were last time and I selected against tipburn. What's going on here? I'm not making any progress. This isn't effective. Well the environment wasn't bringing out all the deleterious traits the first time. So if I really wanted to improve that heirloom lettuce for this trait, then I would actually grow it under conditions that bring on tipburn. I would not put gypsum on the soil. I'd let the surface of the soil dry out. I would overhead water it on hot days. Then maybe my original heirloom would show 50% tipburn. If you get rid of the 50% tipburn that you're seeing under challenging conditions, that's much more efficient. Then the next time you grow the offspring in a normal situation (normal levels of calcium in the soil, you treat it as you would if you're trying to prevent tipburn, playing your normal environmental role as the farmer) on something you've selected in a challenging situation, you may not see any tipburn. The concept here is the challenging environment vs the unchallenging environment.

If you get a lettuce, not an heirloom, that has come from a much more uniform source such as some modern Dutch program which has found their perfect single head of lettuce, blown it up and produced a pure line and you put that out in the garden and you find, under those circumstances that you have 50% tipburn, no amount of selection is going to change that because the genetics are very uniform. Every plant there is really the same, only the micro environment differs from plant to plant. Some of them are in an environmentally challenging situation and are showing tipburn. If its growing on a lump of gypsum where there's plenty of calcium and it is near the sprinkler so the soil's not drying out and the calcium uptake is optimized, it is not going to show tipburn. But there's no genetic difference between them because they were derived from a single plant that was selfed for nine generations. There's no genetic diversity in that lettuce to choose from. You could select that stuff from now to eternity and you would not make any genetic progress. You would find out how good a farmer you were because a plant like that is reflecting its environment pretty accurately. So if you're a good enough farmer that you can make the environment perfect for this plant, you won't get any tipburn. If you put it under a situation where everything is equally challenged maybe you will get a very high rate of tipburn.

Q: If after the first generation you had not selected out the 10% that had tipburn, wouldn't you have had even more than 40% tipburn under the challenging conditions in the next generation?
A: Probably. I selected as efficiently as I could, I took out all the plants that had tipburn but the conditions were not such that I could see all the tipburn-susceptible plants.
Q: Are you advocating doing this?
A: Doing what? Pure line?
Q: This rogueing the 50% of the population because it shows tipburn one year.
A: It depends on the genetic history of the plant. If you're working with a pure-line lettuce
Q: If you knew you had a multi-line
A: Aha, that's different. If I know I had a multi-line
Q: If you think the 50% tipburn is an expression of a genetic group within that line you can throw that out
A; Then you've made progress
Q: Well you've either made progress for tipburn or you've sacrificed something else.
A: Possibly.
J: There comes a time when you make a distinction between 'I'm doing this as preservationist work or I'm doing this to get a better tipburn-resistant lettuce.'
A: It depends upon what your goal was. If your goal was to preserve an heirloom exactly as you received it you wouldn't throw anything out.
Tom: You're definitely losing genetic diversity when you're selecting and hopefully you're losing genetic diversity of stuff like tipburn, bitterness and traits you're not interested in.
A: On the other hand you're going to have some type of crossing taking place that is changing it a little bit.
A: If it was a self-pollinating plant that I was trying to improve for something as serious as tipburn I would probably throw out all of the tipburn.
Q: But let's say you were trying to preserve an heirloom line, then that becomes a more difficult question.
A: It does become a more difficult question but I'm not really a preservationist so I don't have that problem.
Q: You might be wishing to preserve some genetic diversity within your population simply because next year conditions might be different
A: Yes, and that's why
Q: Would you choose ever to keep some undesirable traits because they may be riding along with some unknown desirable traits that you might wish you'd kept?
A: I do that in crossers. I don't do it in selfers. Turns to John: Is that the right answer?
J: Its one of the right answers.
A: In crossers you'd be running a risk to throw out everything that showed one particular trait that you didn't like. You'd have to be assessing: 'Does this have several bad traits? If so, you throw it out. This is what we call getting rid of the uglies in crossers. But you'd keep a little ugliness in your population because associated with that ugliness is something beautiful, the right kind of stem, the right color, the right leaf shape. In crossers, it's a different strategy. You're not throwing out more than 40% of your population at any one time without running the risk of throwing out good stuff with the bad.

In a selfing population, because of the way that they segregate out into these lines, I think that you can throw out 100% of something that's really bad if you have a big enough population. It comes down to the size of the population. If you have 200 lettuce plants you can throw out everything that gets tipburn. If you have only 10 plants then you'd be running more of a risk of throwing out really important genes by throwing out everything.

Biology is never clear-cut. It is a non-linear dynamic system. If we talked about selecting for one particular thing in an heirloom that naturally contains some variation (although overall it all looks the same, in some particular trait that you really hone in on you'll see variation in the heirloom population.) In the pure line population there's not enough genetic variation to effectively select. If you have a plant that's really a pure-line variety or even an heirloom that lacks enough genetic diversity to select from, if you want to have something to select from, you need to induce variation by crossing.

I'll tell you how I got interested in plant breeding. I was growing salad greens and I had Salad Bowl lettuce from seed that I had saved myself. I had grown thousands upon thousands of packets of this salad bowl. One day in the flat there was a red Salad Bowl. I never had studied genetics but I knew that this was important and I thought 'Wow! A red Salad Bowl.' At that time (1983) I had never heard of or seen a red Salad Bowl. So I thought I had something really unique so I saved my red Salad Bowl. I thought that when I saved seed I was going to get Red Salad Bowl the next time. That's not what happened at all.

When I grew the seed for the green salad bowl I was also growing seed for Red Winter Cos. Turns out there was a chance cross between the two that produced my red Salad Bowl, an F1 cross between Green Salad Bowl and Red Rouge D'hiver. The cool thing about this experiment that nature started for me, was that the traits in one are almost completely different from the traits in the other. This is green, that is red, this is oak leaf wavy, that is entire (no serration), in this the head is really open, the leaves are hanging out like this, in that the leaves are standing straight up. These things were obviously different about these two varieties. A genetics teacher couldn't have given me a better example to work with. So I saved one single plant of this F1 and I thought that I was going to get the exact same thing the next time. Well that's not what happened.

Instead I got green leaves that stood straight up, red leaves that laid out flat, I had green oakleafs, I had red oakleafs, I had red-spotted oak leaves, some were green but had red spots on them and I didn't understand that. It turned out that Rouge D'hiver is interesting because it actually has two genes for redness, it has the gene that makes the whole leaf red and it has the gene that makes these red blotches on the leaves. When it is blanched you can see these pink spots on the green. So I didn't just have a red and green, I had red and green and green with red spots. So here is the F2 generation which I've started referring to as the genetic rainbow generation because I could find in the F2 every single possible combination of these two that I could think of. It actually wasn't every single possible combination but it struck me that way. So I numbered each one and I had descriptions for each number. Then I grew those out and I took some of them-some of the combinations were just ugly, some were brownish, some weren't ugly but not exceptional, the red wasn't super red, etc. So I took the seeds of the ones that looked the coolest, the most distinctive to me. Once again there was variation, but the variation was like a little mini-rainbow. This is the F3 generation. More segregation happens here. I remember there was a #11 that I wish I still had. #11 was lime-green, it was upright like a cos, had entire leaves. It was cool because the edges of every leaf were real wavy gravy, even though I didn't see that waviness in either of these, it was restricted to the margin of the leaves. I'm sure it must have come out of the genes here, out of the undulation of Salad Bowl, somehow in this #11 was only expressed in the margin of the leaves. That was such a good lettuce and it disappeared. It was immediately stable. When I grew the seeds of #11 they all came out green, they all came out entire and they came pretty much true. So #11 didn't segregate, it just looked in the F3 generation a lot like the F2 whereas these others were varied, they varied in oakiness, some had oak-shaped leaves, some had entire leaves. As it turns out, oak-shaped leaves are dominant over cos-shaped leaves. And green and red is dominant over green. The reason that this #11 came true was that in this F2 there were a lot of recessive genes that were already paired up. There were no red genes in here, so the green showed through here and when it came down in the F3 generation it was still green because it was a cos-shaped leaf, an entire leaf that being a recessive trait, there were no oak genes in here, when it came down here it was still that way. It was a mystery to me, I had biology in high school and that was about as much as I could remember.

Between then and now I have continued to work with the offspring of this original crop and there are varieties in my catalog on the market now that came from this cross. Oaky Red Splash, Wavy Red Cos. Some of them looked just like #11 except they were red. They had that same waviness on the margin; I really like that.

For the F3 generation, you eliminate some and then you do it again and save seeds again. This is the F4 generation coming out of the F3 generation. You're still getting segregation but the changes are getting smaller and smaller. These all basically look alike. You can tell they all came from the same parent. This one there's still segregation, still cos types popping out, still oak types popping out and then there's this red-spotted gene that will show up in the green ones and in the red ones. The F5 was more uniform. With each generation they are more and more uniform and they all look more and more the same, the variations are more and more subtle. This is the typical pattern of a selfing plant. In a selfer you're losing 50% of the genetic variability with each generation. You're going from 100% heterozygosity to 50% heterozygosity to 25% heterozygosity, to 12.5% heterozygosity. You can't follow all of these, you'd have more work than you can possibly do and so you've got to follow only the most promising lines. As you move down, each generation is getting more and more uniform as you're selecting for traits that keep coming true.
J: The old plant-breeding adage is, "its not what you keep, its what you throw away."

Will: Isn't there a very good chance that you could re-create #11?
F: Yes.
Q: How did this cross happen?
F: It happened on its own, due to an insect, probably some little fly, a syrphid fly or a house fly. These things were near each other in the garden and the cross just happened, courtesy of an insect. There was only one out of thousands, less than .1% crossing in that situation. This was a very young garden with a simplified ecosystem, not much wind, no insectary plants. Under those circumstances I got .1% crossing even though these were close together. When I want to make a cross now I take the two varieties I want to cross and I grow them together on a bed side-by-side. I like to grow 100 plants at a time. I probably would start with 200 and I would make sure that the ones that I'm going to allow to flower have the qualities that I want, so if there's any variability or bad plants, or ones that get downy mildew particularly bad I would throw those out before they start to flower. Once they start to flower they have been through a round of selection already. They grow in 4 ft beds, maybe a red lettuce on one side and a green lettuce on the other, I selected in each of the red and the green, I let them bolt and then I push the flower heads together so that as they're growing up bolting I lean them in toward each other so the flowerheads are really clustered toward the center of the bed. Then I don't worry about it, I let the bees, syrphid flies, do all the work because they work so cheap. After they've flowered I pull them apart so they make their seeds, they fluff out, I don't want the seeds mixing together. I save all the green ones in one bag, all the red ones in another bag, then the next year I plant a 200-cell flat as densely as I dare seed it. I plant the red flat and I plant the green flat and then I start judging them as soon as they come up in the flat and I start selecting. I look at the cotyledens. In each 1" square cell there could be 10 or 15 plants, they put their cotyledens out, you can tell the hybrids right away because they grow bigger, faster. If it's a red-green situation when the first true leaf comes out you'll see a reddish tinge to it at least. You might have to wait till the first true leaf if you can't judge it on pure vigor. I go through cell-by-cell eliminating everything that looks like a self. I've got 200 plants in the flat and I can end up with 20-40 F1s. It is a combination of increasing the chances of crossing by insects by having the heads pushed together, then sowing heavily hoping to find 10% crossing. When you put the heads together like that you get a lot more than .1% crossing. You can hope to get 10% crossing and you certainly get 5% even in lettuce. You don't have to do this with 200 plants. I do it because I like diversity and think there are more chances to find good combinations if you have more parents. I want to have as many opportunities to find best genetics as possible. You could do it with one plant and select the F-1s out of that.

Q: What are your thoughts about doing similar kind of work with a selfer like soybean that has much less showy flowers?
F: You need to know the biology of the plant. You need to know if insects visit the flowers. I don't know about soybeans because I never grow them. I know about Fava Beans, they are basically selfers but you can get a lot of crossing by bumblebees in Favas. What visits the plant, what might transfer pollen? I get the sense that peas are so self-pollinating that you'd have to do a Gregor Mendel deal, go in there, take the flowers apart, remove the pollen-producing parts of the flower before the pollen is ripe, come in there when the stigmatic surface is sticky and put pollen from another pea on. You'd have to do it for peas and I suspect also for soybeans. Anyone know how much crossing happens in soybeans? Not much. I have the sense that soybeans are more selfing than lettuces. The flowers are hidden, first of all.
Q: Are self-pollinating plants mainly that way just because of the visual structure of their anatomy or are there some other biochemical or other mechanisms that make them more self-pollinating?
F: I think it is usually the physical arrangement of the pollen and the stigmatic surface and the timing: whether and when the flowers are open during the fertilization stage. The concept of selfers vs crossers really has a lot to do with the environment they are growing in.

When I found #11 it was fairly late in the spring so I let it go to seed but the ripening season was over while only the very first seeds were ripe. I got only 165 seeds from that plant. I got seeds for only the very inner flowers. I was lucky. If I had found this plant in the next sowing cycle I probably wouldn't have gotten any seeds. I planted 100 and probably the rest of them died in the seed packet. I probably should have done it again. So I think I started out with about 100 plants. A lot of them were of the muddy middle. Salad greens depend heavily on visual appeal, things like high contrast between the color of the midrib and that of the blade of the leaf. Good eye-catching qualities are usually high contrast qualities. A lot of these F2s are of the muddy middle where they are kind of brown, the red isn't very beautiful, the leaves are neither upright nor open, not this nor that and just naturally I gravitated towards the ones which had the highest contrast qualities, the spots, the frilly edges and I tasted them too, but I couldn't taste the difference. But by the time I got down to the F3 and the F4 in this #11 which I think I called Evergreen Lime, it was really sweet and non-bitter even after it bolted. It would have made a great celtuce. It had really good qualities and you could see it right in the F2.

You can begin selecting in the F2, but often it is a mistake. If you have enough time and space you should probably grow all the F2s out to the F3 before you actually start selecting. Things being limited, everything has to start narrowing or you'll get lost in the muddy middle.

Q: If green is recessive and red is dominant, where does brown come from?
F: Modifier genes. There's green and there's red and there's red with modifier genes. The modifiers may make the red more intense by making more pigment, or maybe there's a second pigment constituent that's in the muddy ones. There's obviously more than one kind of red in Rouge D'Hiver. I saw two kinds of red, the red that covers the whole leaf when it's exposed to sunlight and the red splotches that would even show up without sunlight. There are the primary genes for redness, then the modifiers which can make it bright red, brick red, etc. There are other red genes in lettuce that determine where the red is on the leaf. There's a gene that controls when just the tips of the leaves are red, there's another expression where the whole leaf is red.
George M: For colors in all species there are all kinds of modifiers.
F: When you're selecting, other pigment genes and modifiers are making the brown lettuces and the ones in the muddy middle. There's a Brune D'Hiver as well as a Rouge D'Hiver, the Brune is actually a butterhead with a brownish caste to it. Then there's something that's copper-bronze which is different from brown. In the Oaky Red Splash there is a very coppery color in the leaves that is completely different from what you see in Rouge D'Hiver and it has the red spots on it.
Tom: Suppose you grew those same things side by side but never tipped the heads together, left the heads within a foot of each other?
F: Say you'd get 1-5% crossing.
Tom: When you're growing your individual lettuce varieties for seed production, not wanting them to cross, what do you feel is sufficient distance to maintain purity under the conditions of having every bug under the sun in your garden? 25 or 50 feet or 200-300 feet?
F: You can have the beds 10 ft apart and get virtually pure seed. When we grew commercial lots in the field, we bumped the lettuces all up against each other. I did that on purpose to generate crosses but I don't sell that lettuce as either one. I put that lettuce in my mixes. I call those edge rows. I harvest those first and that's a separate lot. I take a nice sample for my future seed breeding purposes and I put the rest in Wild Garden Mix. Likely those are still 99% true. When a person grows that mix out they're probably never gonna see that 1 or 2% and if they do, they might say, "wow, look at that, there's only 1% of that in here, might as well save that" and they'll get a career as a plant breeder!
Q: Do you ever grow something down to where you have a mix still?
F: I sure do. That's also in the wild garden mix. I am doing three mixes this year in lettuce for the first time. But the wild garden mix has always been composed of everything I'm selling, everything I'm breeding with and all the seed that's bred down to here. There are segregating seeds in there and in my description I try to make it clear that's what you're buying. I think that's perfectly legitimate. I'm not a preservationist.

Q: I think it's really great explaining the usefulness of an F1 cross because many times
A: Off-type. They're referred to as off-types.
Q: people call and have the suspicion that they are either inert or sterile or somehow will not procreate at all.
A: Oh No. That's not true. It's not that they won't regenerate, its that they won't come true. The F2 is not going to look like the F1.
Q; I try to make that distinction and allow that they are useful for certain purposes.
A: They are. And in fact, a lot of these F2s did look like the F1s. In this case a lot of the muddy middle looked like the F1s.
Tom: If you were dehybridizing a hybrid you'd select those that looked like it and keep selecting and selecting.
A: And self those, that would work but people don't sell hybrid selfers. They sell hybrid crossers.
Tom: Tomatoes.
A: Tomatoes, right on. What other hybrid selfers are there? Peppers, Eggplant.

For organic agriculture, for creating agro-eco-systems that have some stability in them, we want genetic diversity in the landscape. So if you're working with selfing plants yet you want to get genetic diversity within that self-pollinated population, how do you do it? You create what John referred to as a multi-line. There are multi-lines on the market right now in the catalogs of every company represented here. Forellenschluss also called Freckles also called Trout Back is a lettuce multi-line. How many people have grown Freckles? How many people have noticed that the freckling is different from one plant to another? There's a lot of difference in freckling. That is a multi-line, not a pure line lettuce. You can take the most spotted ones and you can grow those out and you can create a line that is all very speckled. I did and I call it Flashy Trout Back. There I'm taking an heirloom that contains genetic diversity as it is being passed down (it is variable but nobody complains about that) but you can take that and you can create three different things that had three different degrees of speckling that anybody could distinguish.

The cool thing about this though is "what if in that freckling there are genes that are linked to that freckling?" That drought resistance is associated with the plants that have the least freckling (this is purely hypothetical). And cold hardiness is associated with the plants that have the most freckling? In trying to create a visually uniform line if I select the ones that are red freckles I just lost my drought resistance. If I were to select the ones that had minimal freckling I just lost my winter hardiness.

J: Sometimes those associations are not that simple.
F: They usually are not that way.

This is a multi-line, here are the ones that are very freckled, here are the ones that are not freckled, and here in the middle are varying degrees of freckling, these things are moving through time, selfing, occasionally crossing, actually genetic segregation occurs when this one crosses to this one, that would be an F1, and this segregates out the next generation. These occasional crosses keep it mixed up which is why you keep seeing infinite degrees of variation between the most speckled and the least speckled. In this plant you can see the qualities because it is pigment. But imagine that you're not talking about pigments, imagine that you're talking about disease resistance, resistance to aphids, resistance to bolting, resistance to downy mildew, any trait you want. They are occurring in these lines, they are crossing now and then and they are producing novel biotypes. Under normal growing circumstances you won't even notice that this is going on, but if one of these combinations brings together resistance genes for downy mildew, multiple-gene resistance, and this one crossing with this one brings together a number of resistance genes into this one thing and then they segregate out and in the process of these segregating out one of these actually gets more downy mildew resistance genes than any of the rest of them have then, whoa! You get a bad year for downy mildew, a good year for downy mildew to take the alien perspective, and a bad year for lettuce and the whole thing is hammered except for this one line. Suddenly a more disease-resistant population has just taken over and the less resistant ones have been largely eliminated and so in this situation if you're growing enough lettuces at one time and you expose them uniformly to a stress, then you will eliminate all the ones that are not resistant, that are half resistant, that are three-quarters resistant and nature will conserve only the ones that are the very most resistant, or at least the resistant ones will do so much better that you will see it.. You'll look at the whole field, wait till you see my pictures of sclerotinia hammered field tomorrow where 90% of the plants are dead and you see 10% of the plants made it to seed. All around these 10% are dead plants. Its possible those 10% are escapes, just lucky, but if the disease pressure was perfectly uniform across the field then you have eliminated the lucky factor by applying a uniform challenge across the whole field. This is the disease nursery concept. Tomorrow I'm going to talk about using a disease nursery to find disease resistance in lettuce. The disease nursery in a self-pollinated crop reveals these occurrences.