A quick review for those who have not been following our ulluco work: Ulluco is a crop that very rarely sets seed. In fact, there are only two published reports of ulluco seed production: here and at the University of Turku in the early 1990s. There is perhaps one published report of seed being found on ulluco plants in the Andes. To the best of my knowledge, there is nobody else in the world doing sexual breeding with ulluco, although there is a vegetative mutation breeding program in New Zealand. Because so few people are working on the crop, there is very little information available to guide our breeding work. That’s not much of an impediment since I’m not really doing highly technical breeding here, but it is always helpful to know more about the genetics underlying traits of interest.
I have been trying to make some guesses about color genetics in ulluco. This is not easy, although it has one big advantage over most of the other Andean tuber crops in that it is diploid. The usual approach would be to make test crosses between varieties in sufficient number to get reliable ratios of phenotypes in the progeny and then to work backwards to determine the likely genotypes. Unfortunately, it is nearly impossible to make high-confidence controlled crosses with ulluco because the flowers are very small and delicate and less than 1 in 400 attempted crosses produce seed. Also, only about 4% of seeds ever germinate, so producing the kinds of test cross progeny required is impossible at our scale. Even if I spent all of my time making ulluco crosses, I wouldn’t be able to produce enough.
So, we work with what we have. I have been able to collect tuber color data for 72 seedling varieties with known female parentage. The most interesting puzzle is that I have gotten a large number of white varieties when no white varieties existed in the parent pool. Rather than starting from scratch, I decided to look at what is known about the color genetics of ulluco relatives. There is almost nothing known about its closest relatives, but beet isn’t too distant (both are members of the order Caryophyllales) and we know a fair amount about it. Beet also produces similar pigments to ulluco (betalains) and so it isn’t unreasonable to guess that the genetics controlling the biochemical pathways that produce these pigments might be similar. This is a model for red, yellow, and white color inheritance in beets:
Red and yellow are dominant traits and white is recessive. Red is epistatic over yellow and white over both red and yellow. In addition, beet has a gene Bl (blotchy) that, when present in homozygous recessive configuration (blbl), produces irregular coloration. This is the simplest color model for beets; there are more complicated systems that posit multiallelic loci. I am intentionally ignoring those for now, because I have so little data to work with. Simpler is better.
This looks like a pretty good fit. I was particularly struck by the blotchy gene. One characteristic ulluco trait is the red/purple spotting on some varieties. It turns out that blotchy in beets produces the same sort of thing, although in beets it also affects the flesh, not just the skin.
So, at first glance, this looks like a reasonably good match for ulluco, but can it account for the development of white ullucos from colored varieties?
Let’s do some Punnett squares!
First, let’s take the simpler case of yellow ullucos. Since, according to the beet system, these would have to be homozygous recessive for R, we only have to look at Y. A cross between two heterozygous yellow varieties will be Yy x Yy:
Assuming that we have yellow varieties that are heterozygous for y, then we will get a 3:1 segregation between yellow and white phenotypes, 25% white progeny.
Next, let’s do a red x red cross, assuming that both red and the underlying yellow are heterozygous. This one is a little more complicated. This is a two-factor, dihybrid cross of RrYy x RrYy:
We get a phenotype ratio of 9:3:4 red:yellow:white from this cross, or 25% white.
I think you can probably guess how this will turn out, but let’s also do a red x yellow cross, again assuming that both factors are heterozygous. This is a cross of RrYy x rrYy:
We get a phenotype ratio of 6:6:4 red:yellow:white from this cross, 25% white again.
These are the crosses that have the best chances of producing white progeny, other than white x white crosses, but since we did not have any white varieties in our original population, I’m leaving those out for now. This is just a sanity test to see if our results look remotely close. So, here are the parent and progeny phenotypes for all the seedlings that have survived long enough to form tubers:
|Parent Background Color||Total Progeny||Yellow Progeny||Red/Purple Progeny||Orange Progeny||White Progeny|
|Yellow||37||28 (75.7%)||0 (0%)||1 (2.7%)||8 (21.6%)|
|Red/Purple||23||3 (13.0%)||17 (73.9%)||0 (0%)||3 (13.0%)|
|Orange||8||5 (62.5%)||0 (0%)||3 (37.5%)||0 (0%)|
|White||4||0 (0%)||0 (0%)||0 (0%)||4 (100%)|
From yellow parents, we have 21.6% white progeny. From red parents, 13%. We should probably ignore orange because the number of progeny is just too small.
I suspect that this might be the wrong way to do this analysis and that varieties should be classified by whether or not they produce red at all, rather than by background color, but rearranging them doesn’t result in much difference in this case. It is also worth noting that I am not considering possible linkage between these genes at all, even though there is evidence of linkage in beets.
It is dangerous to read too much into such small sample sizes, but these numbers look reasonable to me. I don’t know if the varieties that we started with are homozygous or heterozygous for R and Y. Assuming that we have some of each, then the percentage of white progeny would be less than 25%. The percentage of white progeny from yellow parents suggests that most of our yellows are heterozygous for Y and the percentage of red progeny from red parents suggests that most of our reds are homozygous for R.
If I assume that the beet system is substantially correct for ulluco, which I am now inclined to do, it suggests that I should group varieties in the field by phenotype in most cases, since I want to maintain populations of all colors. Or, at minimum, I should isolate the white varieties from the colored varieties. I don’t know how much insect pollination we get in ulluco. I think very little, but avoiding white contamination seems like a good idea. I don’t want predominantly white ullucos and crossing any white variety to any red or yellow variety that is heterozygous for Y is going to come out 50/50 white to colored. In addition, every variety that doesn’t come out white will be heterozygous for Y, so in the next generation of crosses with white, they will all be 50/50 white to colored. In just a few generations of open pollination, white would overwhelm colored progeny. I’ll bet that the wild progenitor of ulluco had mostly white tubers.
The greatest value of this discovery, if it continues to prove out, will be in having the ability to identify the color genotypes of ulluco varieties. For test crosses, it would be most convenient to have a white ulluco that is also homozygous recessive for red, but because the yy genotype overrides dominant red alleles, we don’t know what the red genotype of any given white variety is. But, we can figure it out! Yellow varieties have to be homozygous recessive for red. So, in any cross between a yellow variety and a white variety, there are only three possible phenotypic ratios: 100% red progeny means that the white parent is RR, 50% red means that the white parent is Rr, and 0% red means that the white parent is rr. The last one is the one that we want. Once we have a variety with the rryy genotype, we can use it to determine the red/yellow genotype of any other variety in just one or two generations of crosses (two generations are required when red is dominant in the variety being tested).
This also might shed a little light on one of the big questions with ulluco: has there been ongoing seed production and incorporation of volunteer varieties into the ulluco crop in the Andes? Seed has rarely, if ever, been observed in the field. In the CIP and PROINPA collections, there is not a single white ulluco. In fact, the descriptors for ulluco, presumably produced by people who have seen just about every variation that ulluco has to offer, don’t even include an option for white skin. Maybe nobody has seen a white ulluco before in modern times. Andean ulluco varieties might all be very old. If volunteers had appeared and been incorporated into the crop with any regularity, there would probably be a large number of white varieties. Of course, it could also be true that white is an unpopular color and that white varieties are culled whenever they appear.
The beet scheme doesn’t account for every color found in ulluco tubers. Ulluco has the additional colors orange and green. I don’t think that orange is seen in beets, but it isn’t a huge leap to assume some sort of combined expression of red and yellow pigments to produce orange ullucos. Green is another matter. It isn’t a very common color in tuber crops. I can’t even begin to guess yet because neither of the green heirloom ullucos in our collection has ever set seed. I don’t even know what class of pigment it might be, so that will require more investigation.
Flesh color also does not match with beets, as skin color does, so there must at least be additional factors involved in ulluco flesh color. Many ulluco varieties have colored skin and white flesh, a combination that I don’t think exists in beets. (Are there white fodder or sugar beets with colored skin? If you know the answer, please leave a comment.) Many beets have red flesh, but I don’t think that there are any ullucos with red flesh. The only ulluco flesh colors that I have seen are white, yellow, orange, and possibly green.
It will take a lot more work to confirm this: more varieties and particularly more controlled crosses, but I am happy with it as a working hypothesis.