How to hybridise Salvias

The basic principle in crossing Salvias is the same as for any plant, the designated female needs to be prevented from being prematurely fertilised and then pollen from the desired male plant needs to be placed on the receptive part of the stigma.

Preventing premature fertilisation requires that any potential pollinators (birds, bees, hawk moths etc) are excluded and the anthers are removed from the flower (i.e. emasculation) before the pollen is ready to be released.  For large flowers like S. mexicana this emasculation can be done the afternoon before the flowers are expected to open.  However, I find that for many species (even S. splendens) it is easier to do the emasculation later in the night when the flower has developed a bit more.  When anthers are removed check that they have not dehisced and discard the flower if they have.   If the intention is only to generate seed from a species that does not normally set seed because of lack of a suitable pollinator, then it is not necessary to emasculate.

The flower (S. fulgens) in the center is at a reasonable stage of development to emasculate.

Best technique will vary with each species,  For S. fulgens the upper petal can be grasped
 and peeled back.  In some species the lower lip also needs to be pulled down.

to reveal the immature stigma and anthers

The anthers are tucked back in the flower

Check that both are removed and have not started to dehisce.

Salvia flowers are reported in the scientific literature to be "protandrous", that is the anthers release pollen before the stigma is receptive.  I have not tested this but just in case I delay pollinating flowers until the afternoon or even the next day.  When the anthers dehisce the pollen mass is held together by pollenkitt so that it does not blow away.  The pollen mass can remain sitting on the anther for days in mild weather provided there are no bees around.  At the peak of summer bees can be extremely active and sometimes I have been forced to collect pollen, from varieties where the anthers are exposed, before sunrise to beat the bees.

The receptive part of the Salvia stigma is the tip of the lower fork.  While the style of many Salvia flowers has luxuriant hairs, applying pollen to the style or stylar hairs does not result in any seed.  I have also not obtained seed from only pollinating the upper fork of the stigma.  In species with coloured styles the receptive part of the stigma is often clear.

S. fulgens flower ready to be pollinated, Stigma forks open.

Apply pollen to tip of lower fork, this is the only receptive part.

Check that there is a small amount of pollen adhering.

Unlike many garden plants, Salvias have a dry stigma.  On plants like lilies and roses the stigma is covered by a nutrient rich fluid that provides an ideal environment for pollen to germinate.  In Salvias the pollen must be recognised and accepted before the female makes moisture available for the pollen grain to then imbibe and germinate.  This is a very selective process and many interspecific crosses fail at this point.

After the pollen grain germinates it develops a pollen tube that grows down the style.  Interactions can take place between the developing pollen tube and style that restrict development of incompatible pollen.  At the base of the style the pollen tube must change direction and find the entrance to the ovary so fertilisation can occur.  The pollen tube is directed by specific chemical signals from the ovary.  The result of all these interactions is that a species can restrict the potential for foreign pollen to produce a seed. 

Species can only exist if their genetic integrity is maintained otherwise there is just one big mixing pot.  Natural barriers to crossing exist where species grown in different ecosystems or countries, have different pollinators or flower at different times of the year.  These barriers are easily overcome in a breeding program or, if species are grown together in a garden hybrid plants may occur naturally.  Many interesting Salvia hybrids have resulted in this way.  However, if species have been isolated for a very long time the changes that have accumulated in their genetics and metabolism may prevent fertilisation or the development of viable seeds.  This situation is referred to as incongruent.  Such species would probably be identified as very different by classical taxonomists and they are also likely to be classified as very different based on DNA patterns.

Where species share the same ecosystem (sympatric) they require specific mechanisms to prevent cross pollination so as to maintain their genetic integrity.  Such species may be taxonomically closely related and have very similar DNA patterns.  They are referred to as incompatible.  Traditional plant breeders have a variety of tricks that can sometimes overcome these specific mechanisms such as using mixtures of pollen (mentor pollen) or bud pollination or grafting the stigma from a male plant onto the female's style.  And there is always the sledgehammer, because these specific mechanisms are rarely 100% effective, just making thousands of pollinations may get a few seeds past the barrier.

For a plant breeder trying to decide what crosses are worth making, taxonomy is a good starting point, but the only proof is to actually make crosses.  Mother Nature always has a few surprises up her sleeve and there will always be exceptions to our scientific expectations.  New taxonomic methods based on analysis of DNA etc have the potential to enhance our understanding of relationships between Salvia species.   However, these methods rely on probability analyses to group species and those groupings may not reflect the relationships in the natural world.  When reading such papers, I first look to see how they have grouped S. greggii and S.microphylla.  Where the distributions of these two species overlap they intercross and their progeny are fully fertile, so from a plant breeder's perspective they are the same species.  New taxonomic analyses that do not reach this conclusion are of no value to a plant breeder.

Sometimes if two species cannot be crossed directly, they may both be compatible with a third species that can then act as a bridge to enable characteristics from all three species to be combined.  

Varieties within a species differ in their fertility.  There is no point trying to make crosses with varieties that are sterile.  If suitable pollinators are present, actual seed set can be used as an indicator of fertility or plants can first be self-pollinated by hand to determine if they are worth using as a parent.  If an anther does not produce sufficient pollen to enable it to dehisce, then the plant is sterile.  I also use a microscope to examine pollen and a stain to test pollen viability. 

Species and varieties also differ in other characteristics that may make them difficult parents.  S. buchananii when used as a female has a propensity for the flower spike to suddenly die before the seed is viable, however, as a male it produces a lot of pollen that adheres well to stigmas.  The S. guaranitica varieties I have tried tend to drop florets before seed is mature. 

Achieving fertilisation and the development of a seed is not the end of the task.  The objective is to have a viable plant and there are a range of additional potential obstacles.

There are three different classes of tissue in a salvia seed.  Although generally referred to as a seed, the fruit of a Salvia is more like a very small nut.  The outer-most layer is maternal and softer, it corresponds say to the flesh of a prune.  Then there is then a thin hard layer, also maternal, that encapsulates the cotyledons and embryo.  These outer layers have the same characteristics, including their contribution to dormancy, as the female parent.  Within this outer shell there is the endosperm and the embryo.  The endosperm has two sets of genes from the female and one set from the male parent while the embryo has half its genetic material from each parent. 

Sometimes fertilisation triggers development of the outer maternal layers but the embryo aborts. This results in what looks like a normal seed but it is empty.  If I have plenty of seed from a cross I dissect some before trying to germinate them.  A more frustrating outcome is when pollination triggers development of a seed but the pollen does not contribute to the embryo.  This may sound strange but it is not rare (for jargonistas; this is termed "pseudogamous apomixis").  I have found this situation to be common in crosses with S. coccinea although it took me a long while to convince myself this was a natural effect and not a consequence of my emasculation technique.

Finally, when a seed has been germinated, the resulting plant may lack vigour due to a lack of congruency in the metabolisms of the two parents.  Sterility is very common in interspecific hybrids since the process to develop the reproductive cells requires paring between chromosomes from the two original parents.  Any differences between the parents are magnified at this stage. 

While sterility is a major problem for a plant breeder, it can be a beneficial attribute in a plant variety.  Flowers tend to drop soon after they are fertilised so sterile plants often retain their flowers for much longer and therefore are horticulturally attractive.  Sterile plants are also 'garden friendly' since their progeny cannot pop up where they are not wanted.