Pedigrees of the Cronin watsonias

The rediscovery of Mendel’s principles of heredity at the beginning of the 20th century inspired a surge of ornamental plant breeding by researchers, commercial nurserymen, and perhaps most importantly by individual gardeners.

John Cronin, Director of the Royal Botanic Gardens in Melbourne from 1909 to 1923, had a personal hobby of experimenting with the improvement of garden flowers. He aimed to demonstrate the application of Mendel’s laws to flower breeding, and encourage gardeners to make their own hybrids. He worked with Dahlia and other genera, but particularly the winter-growing South African watsonias, which he recognised as “everyone’s flower” – easy to grow, attractive, a natural for southern Australian gardens.

In a previous publication I lamented that the exact pedigrees of his Watsonia cultivars were lost with the destruction of his papers after his death in 1923. But now the National Library of Australia has come to the rescue with their wonderful resource of newspaper files at Trove. Cronin was a tireless populariser and communicator, speaking at the evening meetings of horticultural societies around the suburbs of Melbourne and giving interviews to journalists.

In the spring of 1904, while employed by William Guilfoyle at the Botanic Gardens, he crossed a pink Watsonia borbonica with W. borbonica ‘Arderne’s White’. This cross may be represented by the following formula (but note that the order is arbitrary, it is not known which was the pollen parent and which the ovule parent in any of the crosses discussed here):

borbonica × Arderne’s White

He noted that pink flowers were dominant over white in the F1 generation, as has been confirmed by other researchers. In spring 1907 he selected one F1 plant with tall stature, dense branching and large flowers. He crossed this with a purple Watsonia meriana and the widely grown red Watsonia aletroides, and also backcrossed it to W. borbonica ‘Arderne’s White’ to create three lines for further breeding:

1. meriana × (borbonica × Arderne’s White)

2. aletroides × (borbonica × Arderne’s White)

3. Arderne’s White × (borbonica × Arderne’s White)

Cronin’s appointment as Principal of Burnley Horticultural College in 1908 seems to have interrupted this work, and in the following year he succeeded Guilfoyle as Director of the Botanic Gardens. By 1913 he had time to resume his watsonia experiments, and on 20 March sowed seeds from his three 1907 crosses at the Botanic Gardens nursery. Six years is not an inordinately long time to store Watsonia seeds, but there would be some loss in viability which may have unintentionally favoured some genotypes over others. Cronin’s management of the plants was another possible source of selection pressure to produce watsonias adapted to Melbourne gardens: he left the corms in the ground over summer, and gave the plants no fertiliser or watering even though 1913-14 was a drought period.

This generation produced their first flowers in October 1914; Cronin stated that these resembled the 1907 selection in size and colour, and were inbred that year. I interpret this to mean that he produced an F2 generation in each of the three lines by cross-pollinating siblings, since selfing would have produced little or no seed due to incompatibility. Thus,

1. (meriana × (borbonica × Arderne’s White)) × (meriana × (borbonica × Arderne’s White))

2. (aletroides × (borbonica × Arderne’s White)) × (aletroides × (borbonica × Arderne’s White))

3. (Arderne’s White × (borbonica × Arderne’s White)) × (Arderne’s White × (borbonica × Arderne’s White))

Large numbers of these seedlings were raised in the main nursery of the Botanic Gardens. By October 1916 Cronin saw the first flowers of the inbreds, which had a wider range of colours than their parents. Some whites showed up, as would be expected from recombination, including some with flowers of improved size and form compared to the original ‘Arderne’s White’. The watsonias commercially released in the 1920s as the Commonwealth hybrids or “Watsonia Cronini” were selections from this generation.

Line 1 would have produced the many Cronin cultivars with a mixture of characters from W. meriana and W. borbonica. These often have subtle tertiary flower colours due to genes from both species influencing anthocyanin pigment production. Floral bracts are typically well-developed and obtuse, compared to the shorter acute bracts of W. borbonica. Examples include ‘Lilac Towers’, which is the most widely grown Watsonia in Australia today and may be the same as Cronin’s ‘Sydney’, and the one illustrated below which may be his ‘Maitland’.

Line 2 would have yielded flowers with long tubes and small lobes like Watsonia aletroides. The one illustrated here was discussed in a previous post.

Cultivars from line 3 are not interspecific hybrids, but selections within the species Watsonia borbonica and would include Cronin’s improved whites such as this, which may be his ‘Hobart’.

This is the same breeding program that was reported in less detail by Pescott (1926) and Cooke (1998).

It’s significant that Cronin did not use a long breeding program: the cultivars released were no more than three generations away from the original genotypes that had been imported from Africa in the 19th century. As he was working with a perennial that is normally propagated vegetatively, he could stop at the F2 with its fixed heterozygosity. I have bred watsonias four generations on from these and other old cultivars, and can attest that hybrid breakdown soon appears. Some of the resulting plants had interesting extremes of flower shape or colour, many were dwarf or weak in growth, but few were gardenable.

In the spring of 1917 Cronin presented this data to the horticultural correspondent of The Leader, and was lecturing on flower hybridisation to amateur horticultural societies with his new watsonias as exhibits. The following year he gave an interview to The Argus, repeating that his new watsonias were produced by first crossing and then inbreeding on Mendelian lines.



Anon. (1917) Melbourne Botanic Gardens – New colors in flowers – The laws of Mendel. The Leader (Melbourne), Saturday 10 November 1917 pp.13-14.
Anon. (1917) Horticultural society. The Advertiser (Footscray), Saturday 15 December 1917 p.3.
Anon. (1918) Botanic Gardens Experiments. The Argus (Melbourne), no.22,553. Monday 11 November 1918 p. 6.
Cooke, D.A. (1998) Descriptions of three cultivars in Watsonia (Iridaceae) J.Adelaide Bot. Gard. 18: 95-100.
Pescott, E.E. (1926) Bulb Growing in Australia. (Whitcombe & Tombs: Melbourne).


Gazania: diversification vs. speciation

The studies on Gazania by Seranne Howis are a reminder that biodiversity can’t always be divided into discrete species. Speciation may form the kind of clearly articulated branching pattern that cladists like when evolution is driven by new niches becoming available one by one. But the sudden landscape-wide diversification of a clade and its sorting by natural selection into stable entities are separate, and almost contradictory, processes – rather like an explosion and the subsequent settling of the debris. Such explosions of diversity have occurred in the Mediterranean climate zones of south-western Australia and in South Africa, as part of wholesale vegetation changes caused by the cyclical climate changes of the last few million years. Many genera were reduced to small populations in refugia during the dry periods, and rapidly diversified again in the wet periods.

The South African biodiversity hotspot has given the world several genera of ornamental plants that have been evolving in this way. Some of them are real gifts to the plant breeder because in cultivation they function as coenospecies with all their genetic diversity available for use in hybridisation. For example, Watsonia (Iridaceae) has been divided into 52 morphological species that rarely hybridise in the wild as the flowers of sympatric populations have diversified to utilise different pollinators. But they all have the same chromosome number, and can all be interbred in cultivation with the F1 generation often showing hybrid vigour and high fertility. My ‘gut feeling’ from working with Watsonia is that it’s a genus with little evolutionary depth, all the species having similar genetic architecture and closely homologous genes. Gazania is another example, in which the process of speciation may not have proceeded even as far as it has in Watsonia.

Howis showed that Gazania includes seven valid, monophyletic species – each reproductively isolated, with a distinct morphology, habitat and genetic identity – but it has the majority of its diversity in a broad complex where morphological, ecological and genetic variation are only partially correlated. The complex may be called an ochlospecies, defined as a polymorphic species with chaotic infraspecific variation intractable to formal taxonomic treatment. In her 2007 thesis, Howis tentatively called this ochlospecies by the earliest published name, G. rigens (L.)Gaertn. But Howis et al. (2009) take a more conservative course by calling it the K-R complex – possibly because G. rigens is usually applied to the stoloniferous sand-binding forms that are distinct from the others morphologically and ecologically, but not genetically.

Another model for understanding genera like Gazania might be found in Vavilov’s theory of homologous variation, which is closely related to Nabokov’s concept of homopsis. A complex that has diversified only since the Pleistocene is likely to consist of populations with similar functioning genes that determine the morphology and physiology of individual plants. This is quite apart from the fine variations in the four non-coding chloroplast sequences and two nuclear spacers used in Howis’ study. The model predicts that similar traits of morphology and physiology would appear repeatedly in response to the appropriate environmental conditions. Homopsis is a type of homoplasy in which the phenotypic similarities are due to underlying genetic similarity, instead of being due to convergence from more diverse ancestors; this would be the case with the stoloniferous ‘rigens’ forms of Gazania that do not form a genetically coherent entity since natural selection has shaped them from the same gene pool as the rest of the complex.

The many forms of Gazania introduced to Australia, and now feral here, are all within the K-R complex. For practical purposes such as gardening books and legal declaration as weeds, a Latin binomial may be demanded. There are three possibilities:

  1. refer simply to the genus Gazania. As none of the seven distinct species are naturalised here, in practice this would mean the K-R complex.
  2. use the name Gazania rigens (L.)Gaertn. to signify the whole K-R complex. This species epithet has priority under the Code, having been published as Gorteria rigens L. in 1763.
  3. separate G. rigens out as the name of the stoloniferous forms that have been planted for sand stabilisation on our coasts. The earliest valid name for the residue of the K-R complex would then be Gazania rigida (Burm.f.)Roessler. This would be similar to the treatment in the 1986 Flora of South Australia, but with G. rigida replacing the later synonym G. linearis.


Howis, S. (2007) A taxonomic revision of the southern African endemic genus Gazania (Asteraceae) based on morphometric, genetic and phylogeographic data. Ph.D thesis, Rhodes University Botany Department. 293 pp.

Howis, S., Barker N.P. & Mucina, L. (2009) Globally grown, but poorly known: species limits and biogeography of Gazania Gaertn. (Asteraceae) inferred from chloroplast and nuclear DNA sequence data. Taxon 58(3): 871–882.

White-flowered Watsonia

Watsonias are grown as garden ornamentals for their spectacular display of flowers in spring. The wild species introduced from South Africa in the 19th century all had flowers in shades of pink, red, orange or purple. H.H. Arderne’s discovery of a pure white-flowered, acyanic, clone of Watsonia borbonica was big news in 1888, and soon it was a feature display at Kew Gardens. Its popularity and wide dissemination as a garden plant are documented by the numerous articles in gardening magazines of that time. Early 20th century breeders – notably John Cronin in Australia and Luther Burbank in the USA – used Arderne’s White with the species then available in their hybridisation and selection programs.

Arderne’s White

Acyanic Watsonia plants have white flowers with cream anthers. They also have completely green leaves, and stems, because they lack the anthocyanin pigments. On the other hand, wild-type plants have coloured flowers with purple anthers; their leaves may have brown or red pigmentation on the margins and bases.

Some results from my watsonia breeding program from 1997 to 2010 strongly suggested that Arderne’s White and five other acyanic cultivars derived from Watsonia borbonica are due to a single recessive allele, a, with the wild-type allele A allowing the formation of anthocyanins. The same allele was present in some coloured cultivars, as shown by the appearance of roughly 25% acyanic progeny when these were crossed. It also produced recessive acyanic progeny in experimental crosses with Watsonia aletroides and W. meriana. It seems most likely that the a allele was introduced into cultivation via Arderne’s White. But since it is a recessive it must have been present in the gene pool before a rare homozygous plant happened to appear and survive long enough for Mr Arderne to notice it.

All the acyanic Watsonia accessions that I examined have something else in common. Their production of anthocyanins could have been blocked at several different points in the chemical pathway, but in fact they all lack the enzyme chalcone synthase, which is needed to produce all the flavonoid pigments, the yellow chalcones as well as the anthocyanins. This fits the explanation that they all have a mutation at the same A locus that inhibits the production of this particular enzyme. To complicate matters, some temporarily show a pink tinge when the white flowers open in cold weather. The possibility that these have a complementary mutation at a different locus was ruled out by a test cross, which produced only acyanic progeny. If two loci had been involved, only cyanic progeny would have been expected from this cross.

A classic, but all too brief, paper by Horn (1963) mentioned the interesting observation of a 3:5 segregation ratio when he crossed Arderne’s White with three different Watsonia hybrids that had coloured flowers, implying that the acyanic gene was dominant. The lack of further detail makes it difficult to compare his experiment with mine. It may be that one or two of the three hybrids were Aa heterozygotes, giving some acyanic progeny when crossed with the homozygous aa Arderne’s White.

the a allele introduced into another Watsonia line with quite different morphology

Many Watsonia cultivars have been characterised in horticultural literature and retail catalogues simply as ‘white-flowered’. However, a named cultivar of an ornamental perennial is normally a vegetatively propagated clone that has constant horticultural properties due to its constant genotype. Descriptions need to be based on several characters if we want to define unambiguously cultivars with unique genotypes. Descriptions would ideally be supported by a type specimen and molecular data as well. A decade ago there was a lot of talk about cultonomy – the taxonomy of cultivated plants – as a separate subject, and there was still an International Code of Nomenclature for Cultivated Plants separate from the more general International Code of Nomenclature used for plants, fungi and algae. We seem to have gone backwards since then, and I fear that the Procrustean attempt to fit plants created by humans into a system of nomenclature intended for wild taxa will cause a gradual loss of knowledge of cultivars.

Ornamental flower cultivars are selected for the shape and colours of their flowers, and for growth traits that make them ‘gardenable’, i.e. convenient to use in gardens. But other morphological characters that have not been directly selected may be useful indicators to distinguish related cultivars. Among Watsonia cultivars, these characters include the relative length of the style, shape of floral bracts, orientation of stamens and shape of the seed capsule. For example, two genotypes of white-flowered Watsonia that John Cronin produced in the 1920s had become lumped together in the old nursery that preserved them: both are tall slender plants with white funnel-shaped flowers. But one has a style equal equal in length to its stamens, while the other has a much longer style that protrudes from the flower. When the two are grown side by side for a few years, other small differences in flowering time, mean height and maximum leaf width can be noticed.


Cooke, D.A. (2010) Genetics of white-flowered cultivars derived from Watsonia borbonica (Iridaceae). J.Adelaide Bot. Gard. 24: 33-38. Download the PDF here

Horn, W. (1963) Flower colour inheritance in some Iridaceae. Naturwissenschaften 50(15): 527-528.

Red and yellow Chasmanthe floribunda

The South African Chasmanthe floribunda (Salisb.)N.E.Br., is a common garden plant in southern Australia. It could be called a low-care relative of Gladiolus, in the family Iridaceae. The wild-type has an orange-red perianth with some yellow on the perianth tube, a purple inflorescence axis and purple anthers. But some of the plants in cultivation have a completely yellow perianth and yellow anthers, with a green inflorescence axis.


I have examined perianths of both flower types under a light microscope at 200X. The red anthocyanin pigment was dispersed through the cell sap in the red flowers, but could not be found in the yellow flowers; the yellow pigment (presumably a carotenoid) was concentrated in chromoplasts in both red and yellow flowers.

Yellow variant

No plants with intermediate flower colour were observed, although the species regenerates freely from self-sown seed (to the extent of becoming a weed of native vegetation). Because of this clear distinction, the yellow variant had been formally named as C. floribunda var. duckittii by Louisa Bolus (1933).

Apart from its lack of red pigment, the yellow variant is indistinguishable from the wild-type in its morphology. De Vos (1985) had suggested that the portion of the perianth tube below the insertion of the stamens (the hypanthium) was shorter in the yellow flowers but did not cite any measurements. However, flowers that I have measured in suburban Adelaide gardens and roadside populations in the Adelaide hills have hypanthium lengths of 9-11 mm, irrespective of their flower colour.

Experimental crosses
The red pigment is also absent from the plumule of the seedling and etiolated shoots emerging from the corms; they are cream in this variant but reddish-tinted in wild-type plants. This suggested the simple experiment of crossing red- and yellow-flowered plants and scoring their progeny at germination for pigment.

Cross pollination was carried out between a ‘red’ and a ‘yellow’ plant; reciprocal crosses were made to check for any apomixis or accidental self-pollination.

  • Flowers of a ‘red’ were pollinated with pollen from a ‘yellow’
    Result: 30 ‘red’ seedlings, 32 ‘yellow’
  • Flowers of a ‘yellow’ were pollinated with pollen from a ‘red’
    Result: 31 ‘red’ seedlings, 35 ‘yellow’

These results are in accord with a 1:1 ratio (Chi-square 0.281, yielding a probability of 0.595). Four randomly selected plants from each of the ‘red’ and ‘yellow’ seedlings were grown through to flowering; all four scored as red produced wild-type red flowers, all those scored as yellow produced pure yellow. No intermediates between red and yellow were produced in the experiment.

The simplest hypothesis is that a single gene is involved, with red fully dominant over yellow; the red plant used in the cross must therefore have been a heterozygote to give the observed 1:1 ratio in the F1. Selfing of the two plants used in the experiment was not succesful due to self-incompatiblility (which is very common in Iridaceae).

From this I concluded that yellow-flowered C. floribunda plants do not warrant formal taxonomic recognition. They are acyanic (i.e. anthocyanin lacking) variants, analogous to the white-flowered forms of many garden ornamentals that are no longer formally recognised except at the level of cultivars. This experiment was first published in December 1998 on my old website, which is archived here by the Wayback Machine, and the conclusion was also mentioned in the Horticultural Flora of South-Eastern Australia. My only rethink since then has been that there is a bit of variation in the intensity of the red flowers; possibly the more intense ones are homozygous for the wild-type allele and the paler ones with more yellow visible on the tube are the heterozygotes.

Bolus, L. (1933) Plants – new or noteworthy. South African Gardener 23: 46-47.
Cooke, D.A. (2005) Iridaceae. In Spencer, R.D., Horticultural Flora of South-Eastern Australia 5: 196-262.

de Vos, M.P. (1985) Revision of the South African genus Chasmanthe (Iridaceae). South African J. Botany 51: 253-261.