Tomato sex question

I would love to know how this works, could somebody explain?
What I do understand:
A stable variety has basically identical sides to chromosome pairs. So when the male and female parts of one flower fertilise each other, the result is the same as if pollen from one plant fertilises another plant of the same variety.

I also understand that when 2 varieties are crossed, the F1 are all the same as each other, due to all having parents with symmetrical chromosome pairs. But there is variation in F2 because the sex cells then have a randomised selection from either side of the chromosome pairs.

What I do not understand, is the fertilisation process. So here are my questions:

When a tomato flower is fertilised, and then produces seeds, is it one male cell (containing 2 gametes?) which fertilised one ovum, and then that goes on to produce many seeds? Or are there many ovum/female gamete and/or many male gametes involved? If it is just one or two gametes of each sex making all the seeds, then where would the F2 variation come from among the seeds of one fruit?

In an unstable individual, such as an F2, is every pollen grain from a single flower different from each other in terms of the random selection of pieces from one or other side of the diploid chromosome pair? Or is there different variation from flower to flower on that individual?

And then a more practical question, that might be answered by the above being answered, is… about the optimal timing for crossing. In particular, I am wondering if the greatest diversity of crossing 2 unstable parents might actually be in the F1 rather than F2 generation. And, I am wondering whether my thought is correct or not that the best way to create diversity from crossing 3 varieties might be to cross the first 2 and then immediately cross them with the 3rd in the next generation. Or 4 way crosses by crossing 2 pairs and then cross their F1s in the next generation.

1 Like

Basically one pollen grain + one ovum = one seed. So for a tomato, there are lots of pollination events happening, one for each seed produced. For an avocado, only one pollination event as there is only one seed produced.
As to the optimal time to “intervene” and make another cross I’d say that depends on what you want to achieve. If I had 3 tomatoes I liked and wanted to mix them up as much as possible I’d do two-way crosses between each pair, so six crosses in all. This mixes both nuclear and cytoplasmic genetic material.
Hope this helps.


As an aside, a few plants (I think not ones we’re mostly interested in landracing) experience cleavage polyembryony where many seeds are clones of each other, formed after a single fertilization event.


Yes. This happens with at least a few of the fruits we all know, like mango.

1 Like

Wow thanks Ray. Gosh that’s given me another thought, one could just mix up the pollen of 10 varieties and cross them all at once with a mother! 10 crosses in one fruit! I wonder if there is an efficient way of thoroughly mixing fresh pollen without crushing it… Of course it makes sense that note-takers wouldn’t like this method but for landracers it should make good sense! Also I think this is the method used for self incompatible tomatoes in genebanks for mass crosses, and now I better understand why, or at least the added advantage over just the efficiency of time. Makes each individual fruit’s seed more diverse also, which could have advantages especially for low scale seed collection anyway.

And yes, two way crosses sound like the way to go for sure. Regarding your 3 tomatoes you wanted to mix up, to have them really mixed, I suppose you would then want to cross those 6 two-way crosses with each other, in the next generation, perhaps two ways again, so that would make … 30 permutations right? Otherwise with only 6 you’d miss out on all the combinations which mix all 3.

Yuzu gives a lot of polyembryonic seeds (cold hardy citrus). One friend told me actually the main embryo is from sexual reproduction but the others are clones only of the mother. I have no idea if that’s true or not. Give me 4 years and pray for my note keeping to go unhindered and I might be able to give a direct report if my plants survive :grinning:

1 Like

Also, since my maths is not so good, I found this handy tool to help calculate permutations. Very handy so sharing it with others who are into crossing manually!

Yes, I had the same idea awhile back! I haven’t tested it out yet, but I’m thinking that collecting pollen from a whole bunch of fathers, and applying it all to the same ovary at the same time may be a great way to get a bunch of seeds with different fathers out of the same fruit.

I got the idea when I found out that, even in humans, sperm isn’t just first-to-the-egg-gets-to-fertilize-it. In humans, the egg cell actually collects a bunch of sperm out of the early arrivals and then chooses its favorite. So fast sperm that are suboptimal for creating a healthy embryo tend to get discarded, and randomness is less of a factor than we used to assume. I figure if this is true with eggs and sperm, it’s likely to be true with ovules and pollen.

In practice, I’ve already seen it happen accidentally in squash. The F1 generation of my spaghetti zucchinis included some seeds that were clearly purebreed spaghetti squashes, and some seeds that were clearly spaghetti pattypans, all in the same fruit. I didn’t bother to hand-pollinate anything; the bees did both crosses and one generation of inbreeding for me. Neat! That makes it way convenient.


I think there are couple drawbacks to this method. First you will get lots of crosses with same mother so there isn’t as much variety as making crosses with different mothers. Second, which i don’t how much it applies to tomatoes, is that it’s possible that mother plant prefers some pollen over other. I think this was talked about in some other treads before. Mother plant prefers pollen that is genetically more distant from it which is good in one way, but if you want include specific variety then it’s disadvantage if it’s pollen is discarded. I would atleast use as many mothers as possible.

1 Like

In 2017 what I did was find a couple of good tomato mothers with exsertion and daub them with lots of mixed up pollen.

In practice this led to discarding many potential crosses because it was simply unknown which boring red tomato was the pollen donor. I ended up mainly keeping those I could figure out who was the father based on characteristics.

I have already generated some third-generation crosses that have unknown fathers. I know the mother is pretty cool, but it is possible the father could be an unknown boring red domestic. Hard for me to keep replanting something like that when I have a wealth of known crosses (eight last year) to choose from.

Then there needs to be a way to detect the crosses in the first place. Red x Red crosses might sometimes be difficult to even detect!

So my preference to generate genetic diversity and keeping populations heterozygous is by making frequent known crosses. I keep having to remind myself that not all tomatoes are equal in terms of genetic level diversity. Wild tomatoes are by far the repository of that.

1 Like

Sure, but we would never think of this method as being a substitute for different mothers. Nevertheless if you have 20 males you need to cross with 1 mother variety then this has clear pros. Especially in terms of time saving. And space saving. Suppose you have 40 varieties and you want to do 2 way crosses, that’s 1560 hand pollinations you’d need to do. If you wanted to cover most, not guaranteeing all, then you could use this method and make… say 2 crosses per variety, 80 crosses. Or even 4 to increase your chances, and that’s still 160, about 10 times less than the 1560 (and that 1560 isn’t even guaranteed to succeed in each cross right?). If you’re just wanting to do a mass growout for a landrace project, then that method would seem to have huge potential benefits. Even in terms of pollen collection, you could use a hand vacuum device to quickly harvest all the pollen since it doesn’t need to be kept separately, which makes a far far quicker per-flower speed rate, and applying the pollen would be way faster per flower since you only need one jar. The mixing process would take time but that’s a one off. So all in all this could make it way more than 10 times faster. And sure it is not ideal for all needs nor drawback free but it seems to me it could be very good for certain needs especially in the landrace context.

Yeah that does seem a significant point also. Personally for my needs it would not be suitable right now, I will continue on an individual level because I have a small number of crosses to make and want to maximise chances of success for each cross and keep a track of them for at least a couple of generations. - I want to make sure I get F2 seeds from every single cross I have planned, even if it takes me more than one season to make the crosses successfully. (Though I look forward to that stage when I can stop keeping records!) Especially my space for growing is limited which is another reason to keep track, minimise duplicates, and do everything quite deliberately.

Why did you include the pollen of boring tomatoes?

If the mass mixed pollen procedure is done on emasculated flowers, covered from random pollen if need be, does that not remedy this issue?

Yeah that’s what I am attempting. Specific and recorded crosses, specifically focusing on:

  • short strong stemmed early big domestic tomatoes
  • domestic-wild hybrids that are if possible tending towards the above characteristics
  • blight resistance in the above if possible
  • wild tomatoes that can add resistance to wet; dry; disease/pests; and have good smells; and exsertion

Why did you include the pollen of boring tomatoes?

(Chuckle.) My guess is there was something great about the plants other than the color of the fruits. I don’t know about you, but I’d cross boring red tomatoes together on purpose if one was the most productive and one was the earliest to fruit! :wink:

You could save space only if you know what you are working with. Random crosses you might be growing 100 same F1s and you really don’t want to cull F1s either. If I make 100 unique crosses I only need to grow one plant each and there must be same or more genetic diversity in those 100 plants as there are in 1000 randomly crossed plants. Random crosses between F1s would work better as the next generation is not staple anyhow. But then you would be getting less seeds and would need still quite many crosses if you want to make mass grow out. You can always share those F1s you are not using and that way divide the work.

For me, that would make them not boring!

Well in my example, it was 1560 individual regular crosses vs. 80 or 160 on the mixed pollen method. Enough plants for 1560 flowers (not that you’d be able to catch them all in time nor that all would be successful but let’s call it 1560) would take a lot of space right? Lot’s more than 80 or 160.

Sure. And like I said it would not be suitable for every need. But if you’re aiming for 1560 crosses and don’t need to necessarily make/keep all of them, this could be handy. For example if all the ones you’re crossing are already crosses, so you want to get as many of the 1560 next crosses as possible and then next step would be to do a mass growout on a big enough piece of land, this could be really useful.

Yeah that was actually the main application I was thinking of for such a method.

I’m not saying it’s a method that would suit your needs. It doesn’t even suit mine right now! But basically I am seeing it as… perhaps we could consider it as a method in between Joseph’s method and the usual crossing method. Joseph’s is just to plant them out and let the crossing happen naturally, so there’s far less crossing and none of it is recorded. The usual way is completely deliberate and recorded, and extremely time consuming - in these respects it is the opposite of Joseph’s but with maximum crossing results if you have the time and space. The method I proposed would give way way more crosses than Joseph’s, but still be unrecorded. It would take way less time and effort than the normal crossing method but more than Joseph’s method. So I would expect that for some people wanting to make a hybrid swarm for starting a landrace, it could be very suitable, for at least some stage of the process. Or perhaps at times even for people maintaining a landrace that doesn’t have a high enough rate of outcrossing - and in fact that’s exactly what this method is used for by genebanks, right? Maintaining their wild populations in the absence of the accession being isolated by geography and in the presence of wild pollinators - so they have to pollinate them themselves, and use this method to cross them all up, like would naturally happen for self incompatible accessions in the wild. And they do it for both SI and SC accessions.

I meant the the grow out, not the crossing part takes lot more space if you don’t know what you are growing. You would likely have atleast 10x duplicates no matter how you much you cull them. That’s fine if you have plenty of space, but if you want to use smaller spaces then knowing your crosses helps to conserve space. Making those crosses takes more time, but growing initial varieties doesn’t take that much space compared what you need when you get to F2 with those crosses so saving space there isn’t really necessary. In any case you will end up increasing your space as you get futher with crosses. It would be interesting to know what is the actual crossing rate if you use multiple pollen donors. From my own experience I know many tries don’t take at all and many end up having just few seeds. So do you actually get crosses with multiple parents often or does it just take most suited pollen and you end up having 2 way cross that you have no idea who the parent is. When it’s all so random you might not get at all what you are counting on getting. I think in initial stage it’s better to know what you have and what you might still want to include if it’s not succesful in the first year.

Ha ha! Fair point!

As for your idea, I’m sure it’ll perfectly suit some people’s growing tastes. It may even suit mine! I’ve been thinking of doing something like this in order to boost the speed of crossing in my landraces. I also think it’ll be valuable just to get practice at hand-pollinating, even if I see no other benefit from it. It’s a skill I’d quite like to be good at. So it’s sort of a, “Why not? I’ll try it!”

Well, in my example that’s 1 1560 crosses so you’d have to already have quite a bit of space for 1560 plants! But also, if you’re crossing crosses like we both said is probably the most suitable application for this, then they wouldn’t really be 10x duplicates, right? They’d still have variation between them.

But let’s take that space concept - suppose you wanted these 1560 crosses but are crowd sourcing the growing. And that’s a big part of the landrace approach right? Well having 10x duplicates or variants is a good thing in that case, not a bad thing. If you did have 10x then we’re talking about 15,600 seeds to grow out - send them out to the numerous growers and let the selection begin! In such a context the benefits I mentioned still apply and the drawback of space for growing no longer applies.

Also having 10 seeds of one cross can hardly be considered a hindrance of excess if you are weighing it up against regular crosses which use 1 pollen type per cross and thereby get what, like around 100 seeds? But if you wanted to send out seed to 10 growers and really want to keep track of which is which, are you going to process and count out 1560 different seed packets, 10 times over? That’s a lot of work!!! Sure it can be done, but all I’m suggesting is that for some people they might prefer to not do that, instead processing the seeds together then making mixed bags to send to each grower. That is after all what Joseph himself does, right?

My example was for hybrid swarms, not simple crosses grown to F2. If you want to work on each line or grex separately, then sure, this would not be the appropriate method. If you want a hybrid swarm so you can initiate a landrace, this so far as I can see, could be quite appropriate.

That seems a question worthy of experimentation. I would say at least that since this is the method used by genebanks in maintaining the genetic diversity of their seed stocks, this would indicate to me that the experts see it as working. But for sure it would be great to see it tested in a manner where we can see more obvious effects. But there’s something interesting in that perhaps - with domestic tomatoes we may be able to see the effects more easily due to the phenotype differences in offspring, but the actual pollen, there should be way more genetic difference in the pollen of single wild SI accessions than the pollen of various different domestic varieties, so I would assume if that relative pollen fitness thing you proposed is really an issue enough to prevent multiple male parents, that should effect the wild SI species way more than the domestic tomatoes, right? Nevertheless, I think it would be a cool experiment to run. So long as the procedure was sufficient, especially in terms of very thorough pollen mixing to give the method a proper chance of success.

Random doesn’t really mean unexpected. Firstly tests could be run as discussed above, if the method as used by genebanks is in doubt. Second, when dealing with large numbers as in my example, randomness is just a part of the probability calculations.

For example, if your aim is to get ‘heads’ tossing a coin, the outcome is random. If you do it 5 times, it’s random each time. But we can still calculate that you have only a 3.125% chance of failing, over the course of those 5 tosses. So all we need to know is the statistics in order to find a level of acceptable probability of success. So let’s say you can either do 1,000 hours work for 99% chance of total success or 100 hours for 95%, some will chose the latter.

Also by the way I have heard of using mixed pollen as a way of increasing the chance of pollination. I think I even heard this in the context of interspecies crosses. So it might even be that mixed pollen will have increased success rates to be further factored into the equation in weighing up one method against the other.

Cool! It would be awesome if you get the time/energy to track all the seeds from one fruit to see the variation. Well, I guess that would be more obvious if the parents are all stable and you might be working with unstable parents already, in which case would be harder to tell anything from that kind of tracking. But if you do do it with stable parents that would be really cool to track!

Cool! Yeah I’ve been doing a fair amount these days - hoping they’ll take! It’s quite fun doing them :slight_smile: The only mixed pollen I’ve been using though is multiple plants of the sam accession. Right now I need to keep track of what’s what. But at least taking from different individuals should give a lot of genetic variation and also increase chances of successful pollination as I mentioned above, since these are interspecies attempts. I wish you all the best of luck with your crosses!

That wouldn’t be a hybrid swarm in any case. F1s are still staple no matter how many you make them. It’s the F2 grow out that is the hybrid swarm and if you have random crosses you will loose some of that variance before F2 grow out. Having duplicates just dilutes the variance and that’s why you would need so much more space just to get to F2 grow out with same amount of variance. So you might save work with making initial crosses, but then you end up with more work growing same amount of variance. One F1 each is more than enough F2 seeds for yourself and many more that it’s not worth the trouble of getting dublicates to get more seeds, unless you are doing it on a industrial scale. Idea is sure good, but application to this part of crossing doesn’t work as well as you might think. You think about it too much about from crossing perspective than the whole process. Let’s say starting point is that both have 100 unique crosses. Single crosses use more work than random crosses to get, but both have same amount of variance at this point. To grow single crosses can be grown 1 plant each. Random crosses can’t be grown just 1 plant to get all those 100 unique crosses, and so there will be lots of dublicates. How many debends how they are grown. If you grow all the seeds then you will have most dublicates. If you cull you will loose some of the unique crosses, but will still have dublicates and have to grow more than with single crosses. If you grow hundred random F1 crosses then you will likely have at most few dozen unique crosses. I don’t think there is any way you can get more variance with random crosses with overall less work than with making single crosses. I think it’s more than likely that you will be using more actual work to get the same amount of variance. From F2 on it’s whole another story same as with genebanks. At that stage every grain of pollen will make unique cross so it’s more like growing those single crosses. You will need to grow 100 for 100 unique crosses, but with less work in crossing. Random crosses with staple varieties will just have lots of duplicates and one season of growing before you can tap into the genetic diversity. So only the method of pollination is same as in genebanks.

I already covered that Jesse, I will highlight the revelant part of your and my previous comment in bold:

But anyway I do not think this procedure’s benefits are limited to crossing F1s. Making 1560 F1s from 40 parents, sure all F1 seeds of each individual F1 cross will be basically the same as each other, but with your ‘10x’ scenario that’s only 10 seeds of each 1560 varieties which is actually handy not a disadvantage if you’re sending them out to be grown by others, and in fact for some situations this could be an advantage being F1s. For example, if you sent 10 packets out to far off places, that’s around 1560 F1s for each 10 people in a small packet, who can then grow them out on their land and produce the F2 seeds, and it might be worth them keeping 100 seeds of each, making 156,000 F2 seeds to grow out the next season, for example by collaborating with local growers. I can certainly imagine that being an advantage for someone over having to produce those 156,000 different seed types for each 10 groups!

I’m not sure I understand this statement. The way I understand a grex is the variants (e.g. F2, F3 etc.) from two parents. The way I understand hybrid swarm is a whole bunch of different hybrids crossed with each other, i.e. a very varied population originating from many different parents, not just 2. I would usually associate that with many F2s or later generations but still from many different original crosses, but I think if someone gave me a bag of 1560 different F1s I would probably also consider that a hybrid swarm since it is a swarm (“a large number of people or things”) of hybrids. Anyway, since I was specifically focusing my point on 2 way crosses of 40 F2s, I can’t see how anyone would not consider it a hybrid swarm.

It’s up to you whether you consider manually making 1560 crosses ‘industrial scale’ or not :wink: Or, more to the point, it\s up to any individual to weigh their own time and chose which method they want to do. We might all make different choices depending on our needs and preferences!

I’m glad to hear you think it’s a good idea. Also, I did think about it considering not only crossing but also seed processing, seed labelling, seed storing, and seed packing, as well as the number of plants that would be required to be grown to actually make 1560 crosses in the first place in both methods. The method I proposed has benefits all 5 of those criteria.

‘Both’ what/who? And what do you mean by ‘random crosses’?

So I think I might be following what you mean - you have left my 1560 example and gone for a 100 example. I will go with your ‘10x’ example, assuming that you are talking about the method I proposed when you say ‘random’. So you’d have 1000 seeds. Already this is a far smaller number of crosses than I was talking about so yes sure, you as an individual might chose to not use this method for that. But I want to check your assumption of “at most few dozen unique crosses” - if my maths is correct, if you chose 100 seeds from that group of 1000 seeds then on average you will have around 65 unique crosses. Or if instead you chose 200 seeds, then on average 83 unique crosses. My maths might be wrong, you’re welcome to challenge it.

By the way the TGRC recommends this method for self compatible species so not every pollen grain would make a unique cross in such circumstances even in a genebank.

I’m a little tired now. Goodnight!

Like I said your calculations are only based on how much easier the crossing part would be, in theory. There is no reason why you couldn’t make the same amount of unique F1 crosses with same staple variety base by making single crosses, right? When the work evens out in next stages you don’t gain anything by saving work in crossing. Statistically you can’t even get as much variety with random F1 crosses as with single F1 crosses from same base material so it doesn’t give you more variety either. Your math is probably correct for seed selection with even amounts, but there will be uneven amounts and some will be favoured over others. Same as with growing part. Whatever the number is you will always have to grow out much more than with making single crosses. What I meant with industrial scale was referring to seed F2 production, not the cross number. There is now reason you would need to grow lot’s more dublicate F1 to get more F2s to share. One would be enough. And if you were to share F1 mix you would make it harder for those growing out to get variance. F2s and species wide crosses is whole another story so let’s not mix it up. If your main idea was for crossing F1 then use that because it works a lot better for that than making those F1s.