I have read repeatedly that self-incompatibility is more common in species roses (compared with advanced hybrids) and in diploids (compared with tetraploids, etc.) Do we know anything about the genetic cause of these differences in the extent of self-incompatibility?

Note that I’m not asking about how the incompatibility physically works (i.e. in the pistil) but instead about the genetic basis for the differences in self-incompatibility.

I can theorize at least one possible reason for the difference in self-compatibility in different ploidy levels (although my theory might easily be wrong). But I’m not coming up with any compelling theories at all regarding the comparison between species roses and advanced hybrids…unless we presume that there was unconscious selection for self-compatibility under conditions of domestication, which hardly seems (to me) like a compelling explanation in this case. (Also, I believe that a lot of other woody plants–such as fruit trees–have maintained self-incompatibility under domestication.)

Any thoughts?


I’ll start with an aside.
The wild hyacinth (diploid) is self-sterile. Triploid hyacinths are not only fertile, they are self-compatible.

Self-incompatibility is a surprisingly complicate matter, even in the Rosaceae. For example, Heideman (1895) found multiple sexual systems in Prunus americana. Once he became aware of the different systems, he was able to make wider crosses, even between Cerasus avium var. and P. Besseyi,



The information about hyacinths is interesting. So too the information regarding Prunus spp., although as a technical note, I am fairly certain that Cerasus is an old genus designation that is no longer accepted. I believe that Cerasus is now considered as part of Prunus (i.e. Cerasus avium is Prunus avium), so Heideman’s cross would be interspecific but not intergeneric. Nevertheless quite interesting information.

Still interested if anyone understands the genetic mechanism(s) by which self-incompatibility would differ in diploid vs. haploid (etc.) and species vs. advanced hybrids. I do not have any compelling explanation for why, in example, a diploid species might be self-incompatible, but a diploid hybrid self-compatible. This all assumes that the basic information that I have heard (regarding self-compatibility levels) is correct, of course.

Matt Lustig

Re diploid species incompatible but diploid hybrid not being. The two species may have different self fertilisation barriers and neither are dominant/are compatible with each other or are located in different loci allowing the genes to be removed in further generations. Breeders would likely select for those that set seed easily (ease of breeding) which could be an indication of self fertilization if there arent a lot of pollinators around.

Without knowing actual mechanisms of incompatability I can think of some simple reasons for the ploidy effect. Assume that immunity requires recognition of “other”. Therefore it requires recognition of “self” and a way to prevent responding to self in a bad way. That is how it works in humans. Some peculiar situations trick the system so that it mistakes "self’ for “other”. That give an autoimmune disease. During sexual reproduction, the tables must be turned allowing “other” sperm to get to an ovum without rejection.

Self-incompatibility depends on the same idea of recognizing self and other. Only in plants the male and female gametes are (usually) produced on the same plant, commonly even in the same flower. To encourage outbreeding the plant (commonly at the stigma) prohibits self pollen from transmitting sperm to ovum. But now make a tetraploid from two different species (or at least fairly different strains). When pollen and ovum are made, which sets of chromosomes are expressed in each tissue of the flower? Quite possibly some mixture of both. So a lot of pollen will be recognized as non-self compared to the stigma and therefore allowed.

IN nature, if two different strains cross to make a hybrid, the progeny has four choices. Cross with a parent, cross with itself, cross with a different hybrid, or don’t cross at all. If it doesn’t cross with a parent but crosses with itself it effectively becomes a new strain in a few generations. If it crosses with only one parent, it merges with that parent over time. If it crosses with another hybrid, its progeny will segregate our in each generation. If it does none of these, it dies out after the 1st generation.

In brassicas the diploid B rapa (turnip, Chinese cabbage) is self-incompatible under normal conditions. But with strong selection by growing plants in isolation, occasionally there will arise a self-compatible mutation. You can read about this under Wisconsin Fastplants. On the other hand the tetraploid B. napus ( B. rapa x another brassica) is self-compatible. For canola oil producers this is a great boon allowing selection of the very best seedlings, then multiplication of the seed though simple selfing. That way the very uniform seed gives a very uniform seed crop and lots of seed for every flower of every plant. For turnips or cabbages you don’t need a high seed yield, just enough diversity in the population so that bees can cross pollinate the flowers of plants growing close together.

Species roses in nature are generally growing close enough to get pollinated by a different “non-self”. Only human breeders care to increase the seed yield. High seed yield is no big advantage if asexual reproduction is effective. So our prairie rose produces few seeds per hip and lives on in the prairie vegetatively through a multitude of fires. Species like multiflora that don’t sucker make zillions of seeds and attractive fruits that birds use to assist distribution.

I would guess that for multiflora, maintaining a very large germplasm is important to adaptation to changing conditions and varied ecosystems. So crossing with other plants is an advantage. For prairie rose it is less important. It survives where prairie does. You can make up rationales for most everthing. What’s hard is testing them.

Here is a paper that may interest you:
Botanical Gazette 73:110-132 (1922)
A. B. Stout

As in many other cases, self-incompatibility is not absolute. Sometimes environmental conditions (e.g., temperature) may weaken the incompatibility. In the examples Stout studied, the shift in incompatibility occurred over the growing season. “…it was noted that the highest degree of self-compatibility attained by any given plant appeared very uniformly during the period of mid-bloom.”

There are also some examples of plants that are more willing to form hybrids early or late in the bloom season, or when the plants are young and flowering for the first time.

Stout’s paper contains a particularly interesting and thought-provoking discussion:

“It is to be noted that the complete life cycle of flowering plants involves two periods of vegetative vigor and maturity; one for the sporophyte and one for the gametophyte. The former culminates in the production of spores and the latter in the production of gametes. The generations are antithetic. In its length of life, vigor of vegetative growth, and reproductive power (number of gametes), the gametophytic phase has become relatively weak and highly specialized. In the sporophyte great vegetative vigor is correlated with great reproductive vigor in the production of spores (which are, however, in themselves asexual) and in the nurture of the gametophyte and the embryo. Sex differentiation in the great group of flowering plants has been pushed back during the progress of evolution into the sporophytic stage of the entire cycle, and here sexuality now culminates in seed formation in which the nutrition of the embryo is a most important factor. Sexual reproduction in these higher plants has become more and more inter-related with the vegetative phase of the sporophyte and subject to its internal and biogenetic regulation.”
Emphasis added.

For many years I had puzzled over the apparent reversal in the durations of the alternating generations. In mosses, for example, the sporophytes are small things of short duration, while the gametophytes are generally much larger and longer-lived. But in the angiosperms the gametophytes (pollen tubes and egg sacs) are tiny, whereas the sporophytes may live for decades and more.

I wondered about the reversal, but had not considered that I had hold of the wrong end of the stick.

Could “hybrid vigor” have played a role? Monoploid mosses cannot exploit hybrid vigor, and the diploid moss sporophytes don’t live long enough to get much use from it. But what I may be reading into Stout’s paragraph is the idea that the gametophytic function (or program) began to run on the sporophyte’s diploid chromosomal computer (so to speak). The physical gametophytes became reduced to almost nothing (still necessary, of course), as their sexual functions became parasitic on the sporophyte.

There is the matter of incongruity. I do it my way, you do it yours. If our methods are compatible, we are ahead of the game. If not, nothing gets done.

Two diploid Rosa species may be closely related, like R. woodsii and R. blanda, and likely to share similar incompatibility mechanisms. Or, they may be as remote as R. setigera and R. macrophylla.

The S-allele approach is apparently wide-spread, and found in plants that are not at all closely related. There is another odd factor sometimes found in these plants: a separate gene that turns the system upside down.

In the common system, a diploid S2S5 would not accept pollen of the S2 or S5 types, but would happily accept S1, S3, S4, etc. But if the plant also inherited the F (fertility) gene, then it would happily accept S2 and S5 pollen, but not the other types.

And so it is not difficult for me to imagine two distant species that have different incompatibility systems. One relies on the S-allele system, while the other depends on heterostyly for self-incompatibility, but carries the F gene. The incongruent incompatibility systems nullify each other.

Yes Karl, you are right in that last post. I didn’t want to get into details in my post above. There are too many incompatibility systems to deal with, and I don’t know just how roses do it. For Brassicas, pollination prior to flower opening overcomes the SI if you really want essentially homozygous selfs, or to cross SI pairs. But in natural environments that’s not an option. And yes it’s leaky because a few % of oddities is always a good (small) investment for the plant. Just like seed banking which wastes reproductive potential in an agricultural system, but is really beneficial for opportunists such as weeds. R canina behaves that way. So do other species to a greater or lesser extent as I’m gradually showing year by year in germination studies. Of course van Fleet and others knew that a hundred years ago.

I was going through my data base of useful information and found one that is precisely on point:
Elements of Genetics pp. 2455-256 (1950)
Darlington and Mathers
Now, we may ask, what happens when we cross two self-incompatible species, each with the S genes? This has been done for Nicotiana alata and N. forgetiana in the production of the garden form N. sanderae. Pseudo-compatibility, that is successful fertilization with pollen having an S allelemorph the same as in the style and therefore not legitimately capable of growth, is unknown in forgetiana. It occurs only rarely in alata. In their derivative it is common, especially with certain of the weaker allelemorphs. Thus the recombination between the general gene systems of two species has robbed each of its efficiency as a basis for the action of the S series.

The S-alleles appear to be in charge of the incompatibility system, but they are only the switches (so to speak). Other genes are responsible for maintenance of the system as a whole. Come to think of it, this system is rather like the inheritance of Batesian mimicry in butterflies that I discussed in another thread.


you guys complicate these things to much, going into great lengths on definitions and theories. Just simplified it as, it can’t pollinate itself due to a hormone block, but a plant of the same species, growing nearby because it is individually different, can pollinate and vise versa. A lot of plants do this, maybe its a way of protecting its self from inbreeding and accelerating faults. Its like when some one made the comment , why modern roses are mainly tetraploid. Most of the garden roses in Europe before the Asiatic influence were tetraploids.

It is so much easier to describe a thing when there are no facts to get in the way. The self-incompatibility mechanism involves the coordinated action of multiple genes. The S-alleles are switches.

In addition, some incompatible species have (or can mutate) a fertility allele: Sf. This upsets the system, allowing the carrier to self-pollinate or cross-pollinate. I don’t know for sure, but I do recall that Sam McGredy V commented (back in the 1990s) that his ‘Sexy Rexy’ had to be very carefully emasculated because the variety accepts nearly every sort of pollen that is thrown at it, including its own.

Another fact that should be considered is that a specimen is not a species. If this one won’t accept its own pollen, we cannot assume that no other member of the species will be as stubborn.

Finally, incompatibility and cross-compatibility sometimes may be disrupted by environmental conditions (soil, temperature, rootstock). E.g., Darwin reported that Eschscholtzia californica was perfectly self-sterile in Brazil, but the descendants of those plants were partially self-fertile in England.

This is what happens when one doesn’t proofread his own post carefully enough. This obviously should have been “diploid vs. tetraploid” not “diploid vs. haploid.” Oops!

This is certainly correct regarding the possibility of a hybrid crossing with a parent (aka “backcrossing”). Thus we sometimes have traits that appear to have moved from one species into another through introgression.

However, I believe that the distinction between a hybrid individual “cross[ing] with itself” and “cross[ing] with a different hybrid [individual]” is overstated.

Although it is true that a hybrid that crosses with itself could become “a new strain” (at least under selection pressure, whether natural or artificial) so too this could occur in a population of hybrids between 2 parental species. Thus the same outcome could occur whether the individual hybrid crosses with itself or crosses with (other) nearby hybrids.

It is true that if a number of hybrid individuals reproduce together, their (F2, F3…) offspring may show traits essentially anywhere along the continuum between the parental species. However, assuming heterozygosity, the offspring of an individual hybrid that reproduces with itself could also show traits along that continuum.

I was using a sort of shorthand when I mentioned “crosses with itself”. I intend it as a logical possibility to compare with the cross to a different hybrid (that is one with different parents). If the species has strong self-incompatibility then of course this logical possibility of selfing is skipped.

If you have a sparse population across a landscape and repeated selfing of the single progeny of a one-time outcross, you will get a whole series of similar lines. This is what a wheat breeder does in single seed descent breeding. They make one (fairly) wide cross, grow the seed, let it self, then grow out the progeny of that head sufficiently isolated so that each one pollinates only itself. Then a single seed of the each good looking line is grown, and the cycle repeats for half a dozen generations. Eventually you end up with uniform new named cultivars with a desirable combination of traits. And a ton of discarded seed. We have more powerful breeding techniques these days, but nature may not. Of course there is some segregation in each generation, but the selection pressure of only a few seeds being carried forward in each generation reduces the segregation possible in each generation. In a prairie only a very small fraction of all seeds germinates, even far fewer grow to reproductive maturity. One old line must die for each new one that appears in a stead state system. Agriculture or horticulture are very much non-steady state operations.

Contrast that “crosses with self” with what you get if you put Rosa virginiana x Rosa laxa as female parent and Bayse’s legacy as male parent. What will the OP selfs of a population grown from that cross look like? Huge range of segregation and diversity.

We’re getting a bit far afield from the original topic of self-incompatibility :slight_smile: , but I think we agree on the central points. I think we both agree, for instance, that the same possibilities exist whether a single hybrid individual selfs (assuming a high degree of heterozygosity in that individual, probably a safe bet for a hybrid) or a number of hybrid individuals cross together. The difference in outcome between your two examples (wheat vs. rose) is essentially due to the extreme difference in the level of selective pressure that is being exerted, not really the number of F1 individuals involved (using the term F1 a bit loosely here).

In your wheat example, through repeated episodes of extreme selection (to one seed per line) and selfing, the breeders would be exerting extreme selective pressure and promoting homozygosity, ending up with an extremely inbred and uniform line. The original F1s probably each had a fairly high level of heterozygosity, but with each generation that level would be likely to decrease precipitously.

If the breeders were to “make one (fairly) wide cross, grow the seed, let it self” (same starting point as your example) but then exert no selective pressure but instead allow for an expanding population (avoiding loss of alleles), the level of segregation and diversity would, I think, be expected to be very similar to that in your rose example.

In the absence of strong selective pressure and in a large population of F2s, no doubt true.

Thanks to all involved for the interesting thoughts regarding the differences in self-compatibility levels. It sounds like there is not certainty about the genetic mechanism(s) that would result in these different levels in Rosa specifically, but there are some interesting theories. I particularly like the thoughts about higher ploidy levels making rejection of “self” pollen less effective. It would be interesting to know if self-compatibility is more common in tetraploid species as well, or just in tetraploid hybrids.

Erlanson (1963) addressed self-compatibility in wild roses. I haven’t sorted all the species by ploidy, but here are some diploids:
Foliolosa and Nitida (strongly selfing), Palustris (most strongly, some weakly selfing), Blanda and Woodsii (strongly non-selfing).

It is odd, though, that she found Rosa setigera to be strongly selfing. She also listed R. pimpinellifolia as strongly non-selfing, though Heslop-Harrison (1921) found R. spinosissima to be strongly self-pollinating in his garden. They may have been writing of different species.

I think it is a pity that low altitude Scots Roses (R. spinosissima) are so often confounded with the alpine R. pimpinellifolia.

Interesting that R. nitida is strongly selfing. I pollinated a lot of opened blossoms last year, kind of assuming that self-incompatibility would help counteract my sloppy emasculation. I was impressed by how readily nitida seemed to accept foreign pollen.

R. foliolosa, too, was very accepting of modern pollen.

Kim Rupert’s Orantida is a good example of nitida accepting foreign pollen and Orantida is proving to be very fertile as it self-pollinates and accepts foreign pollen readily.

“It is odd, though, that she found Rosa setigera to be strongly selfing.”
A strongly selfing dioecious species? That’s a pretty neat trick. Tends to make one question the other findings in that paper, however.