Hybridisation in the genus Saxifraga has long been of interest to growers, partly because of the fun of seeing what the result might be and partly because of the anticipation and excitement of producing a desirable new cultivar. With few exceptions, however, there has been little systematic investigation of the breeding limits of particular species across the genus as a whole. This point is important because a knowledge of the extent to which species will cross with each other can tell us a lot about their evolutionary relationships.The ability to cross or not with other species can often be used as a measure of evolutionary divergence. Thus, if two species hybridise and produce fertile offspring, we can conclude that they are probably closely related; the production of sterile offspring suggests they may be less close. The failure to produce any offspring at all, i.e. full interspecific incompatibility, may indicate a much greater divergence and a much less close relationship. I emphasise the word "may" in these statements because the correlation between reproductive isolation and our largely morphological concept of relatedness as expressed in our classifications is not perfect. Thus some species that are believed to be very closely related because they look similar may nonetheless prove to be cross-incompatible. Confused? Don't be. In a nutshell, if two species cross, they are probably closely related; if they don't, they may or may not be related.


In 1926, Murray Hornibrook published an article in the 'Journal of the Royal Horticultural Society' in which he gave a valuable account of the chief points in the history and progress made in hybridising saxifrages up to that date. Comparison of our present knowledge about hybrids with that discussed by Hornibrook reveals that we have advanced only a little in our understanding of breeding relationships across the genus Saxifraga as a whole. Thus today, as in 1926, we can conclude that hybrids are relatively easily formed between species within the same subsection or section, whereas hybridisations between species of different sections are much rarer and frequently subject to some doubt (Fig. 1). It is in this latter category that most work needs to be done. As Fig. 1 shows, there is vast scope not only for checking earlier claims but also for conducting novel inter-sectional and inter-subsectional crosses. Although there is much less need to try crosses within subsections, even here there is scope for further work, particularly on the behaviour of S. caesia and S. retusa, and on the fertility of hybrids resulting from crosses between series.


hybridisation network
 bulletb intra-sectional or intra-subsectional hybrids reliably reported;
 bullet no hybrids reported;
___ inter-sectional or inter-subsectional hybrids reliably reported;
..... inter-sectional or inter-subsectional hybrids reported but dubious;
  no connecting line means no hybrids reported.
 Sections and subsections are abbreviated as follows: 
AIZ subsection Aizoonia; 
CIL section Ciliatae;
COT section Cotylea;
CLN subsection Cuneifoliatae;
CYM section Cymbalaria;
ENG subsection Engleria;
FLO subsection Florulentae;
GYM section Gymnopera;
HET section Heterisia;
HOL subsection Holophyllae;
IRR section Irregulares;
KAB subsection Kabschia;
MER section Merkianae;
MES section Mesogyne;
MIC subsection Micranthes;
MUT subsection Mutatae;
ODO section Odontophyllae;
OPP subsection Oppositifoliae;
ROT subsection Rotundifoliatae;
SAX subsection Saxifraga;
STE subsection Stellares;
TRA section Trachyphyllum;
TRD subsection Tridactylites;
TRP subsection Triplinervium;
XAN section Xanthizoon.
  • Fig. 1
Outline of a project

In order to fill in some of the gaps in our knowledge of breeding behaviour, it strikes me that members of the Society are uniquely placed to contribute. Thus although individually we have limited resources, collectively we have access to as wide a range of species as may be found anywhere. By pooling our experience and results we could amass a valuable set of data which would be of great biological interest. Note that the project would concentrate on species rather than cultivars.

If you would be interested in taking part in this study, then I should be pleased to hear from you and shall offer any further advice that may be necessary. Below I outline the general procedures that may be followed if data of quality are to be obtained. Don't be put off by the sound of it all - it's really quite easy! And remember, even the most limited of crossing experiments will contribute worthwhile information.

Artificial hybridisation

The procedure for doing artificial hybridisations is straightforward, although the ease with which it may be accomplished will depend on the species concerned, and in particular on the size of the flowers. You will need the following equipment: fine forceps, alcohol, tea-bags (or similar), lens, jewellers' tags and a notebook. The steps are as follows:

a) Select the flowers that are to receive the pollen and remove the stamens (emasculation) with fine forceps before any pollen has been shed. Wash forceps in alcohol between each emasculation.

b) Protect the emasculated flower from contaminant pollen by enclosing it in a bag; empty tea-bags with one side opened are ideal, or you can use hand-made muslin bags, or even paper if it can be kept dry.

c) When the stigmas of the emasculated flowers are receptive (in Saxifraga they become moist and sometimes expanded), remove a freshly-dehisced stamen, with forceps, from the selected pollen-donor and rub it gently on the stigmatic surface of the pollen-recipient. Wash the forceps in alcohol between each pollination. A fine 'camelhair' paintbrush may also be used to transfer pollen, but it too must be washed between pollinations.

d) Re-bag the pollinated flower, and attach a label to signify what cross has been made. A jewellers' tag is often suitable for the latter purpose.

e) Record in a notebook what you have done, noting the date, the individual plants used, and the direction of the cross (i.e. which plant was the pollen donor and which the recipient).

One problem you may face is that pollen may not be produced at the time when it is needed. This can be insuperable, as pollen loses its viability after days or weeks, especially if it gets wet, and some lives only a few hours. Generally, pollen grains with two nuclei live longer than those with three. In Saxifraga all species have binucleate pollen, except those belonging to the section Ciliatae, which have three nuclei, and so storage of pollen of the former for later use might be possible. One way of doing this is to collect the anthers just before dehiscence and store them in a desiccated condition (in a capped vial with silica gel). Stored like this in a refigerator the pollen might last about a month and in a deep freeze for about a year, although I do not know of any published experiments that have been conducted on Saxifraga, and species are likely to vary one from another.

Another point relates to whether it is better to grow experimental plants, and conduct the crosses, in a greenhouse or whether to do it outside. Certainly it is easier to manipulate the plants and bag the flowers when they are on a stage in a greenhouse, although I believe that some people have had the experience of better success with their crosses if they have been made out of doors. If you do it outside you must make doubly sure that the protective bags over the flowers function properly or stray pollen will ruin the experiments.

Experimental design

A cross with a positive result, i.e. the production of hybrid seed, obviously tells us something about the relationship of the parents. A negative result, however, is much harder to interpret. The failure to hybridise could be due to adverse environmental conditions (even if only transient), poor health of the plants, wrong developmental stage of the flowers or a number of other causes. To cheek for these, it is a good idea to select at least two flowers on a plant, one for the test cross and one for the control cross. Ideally, the two pollinations should be run simultaneously, although this is not always possible and not too serious if not. In the test cross you pollinate with pollen from the target species. In the control you use pollen from the same plant you are pollinating (if self-compatible) or from another genotype of the same species. Failure of both pollinations would suggest an environmental cause because the control would normally be expected to succeed if the plants are in good health and conditions. Failure of the test cross but success of the control would mean that the cause is likely to be genetic, an important result with possible implications about the relatedness of the two species.

A further refinement to the experiment is to conduct each cross on a reciprocal basis. For example, if you were crossing S.aizoides with S.oppositifolia, you should use S.aizoides both as a pollen recipient (as a female) and as a pollen donor (as a male); in so doing you would also be ensuring that S.oppositifolia was being used both as male parent and as a female one. This refinement is because the success of a cross may sometimes depend on its direction, and this information can be useful.

Measuring success

There are two chief ways of measuring the success of a cross: either count the number of seed-bearing capsules, or count the number of seeds produced in each capsule. Although the latter is more precise, there are significant advantages to the former, especially in terms of ease and speed. Either way, the results should be expressed with reference to the success of the controls. Thus the overall success could be the average number of seed-bearing capsules per test pollination expressed as a percentage of the average number of seed-bearing capsules produced per control pollination.

Information from hybrids

Hybridisation between plants of the same species usually produces fully fertile offspring. Hybrids between different species, however, often suffer various degrees of reduced fertility, and the extent of this reduction can sometimes tell us something about the relatedness of the parents. The simplest way to assess the fertility of a hybrid is to examine its pollen. Pollen viability in living plants may be estimated by mounting the grains in a drop of glycerine on a glass slide and examining them under a microscope. Inviable grains vary greatly in size and appear shrunken or otherwise malformed, whereas viable pollen is plump and of a more or less uniform size. The percentage of ‘good’ grains in a representative sample (of say 300 grains) gives a reasonable indication of fertility. The ability to use a hybrid in further cross-pollinations may also provide a guide to its fertility.

It is also a common misconception that plants with different chromosome numbers will not hybridise. This is not necessarily true because the chromosome number itself does not have a direct bearing on the success of a particular cross. It can, however, affect the fertility of any resulting hybrid.

Anyone wishing to take part in such a project or with results of interest should write to: Dr Richard Gornall, Department of Botany, University of Leicester. University Road, Leicester LE 1 7RH.