Index | Search | Home | Table of Contents

Knapp, S.J. 1993. Breakthroughs towards the domestication of Cuphea. p. 372-379. In: J. Janick and J.E. Simon (eds.), New crops. Wiley, New York.

Breakthroughs Towards the Domestication of Cuphea*

Steven J. Knapp

  7. Table 1
  8. Table 2
  9. Fig. 1
  10. Fig. 2

Over three decades ago, several species of Cuphea of the Lythraceae (Graham 1988) were found to be rich sources of medium-chain fatty acids (MCFAS) (Earle et al. 1960; Miller et al. 1964). These fatty acids and medium-chain triglycerides (MCTS) are used worldwide as industrial feedstocks and foods. They are commercially supplied by the tropical oil crops coconut (Cocos nucifera L.) and oil palm (Elaeis guineensis Jacq.); no temperate oilseed crops supply these lipids (Ignacio 1985; Arkcoll 1988). Many of the MCT-rich species of Cuphea are summer annuals which, if domesticated, could become domestic sources of MCTs for the United States and other importing countries, thereby breaking the tropical oilseed monopoly. Toward this end, a project was undertaken by G. Roebbelen and his colleagues at Georg August University in Goettingen, Germany to evaluate different species and to determine the feasibility of domesticating Cuphea (Hirsinger 1980; Hirsinger and Roebbelen 1980; Hirsinger and Knowles 1984; Hirsinger 1985; Roebbelen and von Witzke 1989). A project to domesticate Cuphea is currently underway at Oregon State University (OSU) (Knapp 1990).

Several species native to Mexico are adapted to the United States and Europe (Hirsinger and Roebbelen 1980; Hirsinger and Knowles 1984; Hirsinger 1985; Knapp 1990). At OSU, we work extensively with C. viscosissima, the only species native to the United States, and C. lanceolata, a species native to the Sierra Madre of Mexico (Graham 1988). These species are grown as summer annuals in the United States. The seed and oil yields of these species seem to be sufficient for them to compete as oilseed crops (Knapp 1990b); however, this must be substantiated using non-shattering cultivars, which are presently being developed. Seed shattering and seed dormancy are the major domestication barriers within the genus (Knapp 1990a,b); the seed dispersal mechanism is a characteristic which unifies the genus (Graham 1988). Only a fraction of the total seed yield of Cuphea is recovered by combine harvesting, the remainder being lost to seed shattering. This trait must be eliminated to domesticate and commercialize Cuphea.

Seed shattering has frustrated the endeavor to domesticate Cuphea. This problem has persisted because no natural diversity has been found for seed shattering (Graham 1988, 1989; Roebbelen and von Witzke 1989; Knapp 1990b); however, over the last three years, important breakthroughs have been made at OSU towards eliminating seed shattering and seed dormancy by exploiting interspecific diversity. These breakthroughs have significantly brightened the commercial outlook for Cuphea. In this paper, I report the discovery of non-shattering phenotypes within the C. viscosissima x C. lanceolata f. silenoides population VL-119, the only non-shattering phenotypes known within the genus, and review the development of autofertile non-dormant and non-shattering C. viscosissima x C. lanceolata f. silenoides germplasm and cultivars at OSU. The development of autofertile non-dormant germplasm has been enormously important to the development of Cuphea as a crop (Knapp 1990a,b), but this has been overshadowed by the genesis of interspecific diversity for seed retention. This has greatly increased the commercial promise of Cuphea by creating the basis for eliminating the most serious barrier to the domestication of this crop.


Throughout the history of agriculture, non-shattering and non-dormant populations have arisen within numerous species, thereby leading to their domestication (Renfrew 1969; Ladizinsky 1985; Kadkol et al. 1984a,b, 1989). The discovery of non-shattering populations of seed crops was critical to the shift from hunting and gathering to agrarian cultures (Renfrew 1969). The non-shattering phenotypes of many diploid seed crops have been shown to be caused by single gene mutations--events which have overwhelmingly affected human history and culture (Renfrew 1969; Ladizinsky 1985; Kadkol et al. 1984a,b, 1989). Natural selection has maximized seed dispersal within Cuphea, thus optimizing its survival in the wild. Cuphea has not been subjected to selection against seed shattering over the last several millennium because it has been of no use to humans as a seed crop. The economic value of this genus as an oilseed only became known late in this century. Domesticating Cuphea means reversing 12 to 14 million years of evolution to perfect seed dispersal, yet, as history teaches us, this reversal might be achieved by a simple mutation.

Until 1987, Cuphea was a completely undomesticated genus. When we initiated domestication work in 1986, the entire germplasm base was comprised of wild populations, and no species had been selected for intense breeding and domestication work. Seed of most species was very scarce, and no species had been extensively collected from the wild for germplasm preservation and plant breeding work. Despite this, our initial goal was to rapidly narrow the field of species down to the one or two with the greatest promise as oilseed crops.

Between 1986 and 1988, we tested many of the more promising species evaluated by Hirsinger and Knowles (1984), Hirsinger (1985), and Roebbelen and von Witzke (1989). From this group, which was entirely made up of Section Heterodon species, we selected C. lanceolata and C. viscosissima for intense domestication work (Table 1). Their selection was partly motivated by the discovery of fertile interspecific hybrids between them (Table 1).

In 1987, Anson Thompson (USDA/ARS, Phoenix, Arizona) reported fertile F1 progeny from C. viscosissima x C. lanceolata matings (Ronis et al. 1990). We subsequently discovered thousands of C. viscosissima x C. lanceolata seedlimgs within a C. lanceolata population. That arose from natural outcrossing between these species. We began using C. viscosissima x C. lanceolata matings for breeding work because C. lanceolata is an extremely useful source of diversity for many important traits, e.g., seed dormancy. We had been selecting, for example, for increased oil percentage within each species separately, but shifted to selecting for increased oil percentage within the C. viscosissima x C. lanceolata population VL-119. This population was created by intermating C. viscosissima and C. lanceolata progeny selected for increased oil percentage (Fig. 2).

Because chiasma frequencies within C. viscosissima x C. lanceolata populations are less than those within either species (Brandt and Knapp 1992), we intermated VL-119 for an additional generation without selectionas C. viscosissima x C. lanceolata populations are allogamous. The additional intermating proved to greatly increase recombination between the genomes of these species and led to phenotypes transgressing the range of these species (Fig. 2). Non-shattering phenotypes (Fig. 1) were among the transgressive segregates which sprang from this population (Fig. 2). At least 15 out of 500 spaced plants of VL-119 had some seed retention. Many of the 15 show great promise and are the focus of intense breeding work.

Wildtype Cuphea disperses seed through a dorsal abscission layer along the corona tube. The placenta of the wildtype rotates upward 100 to 120° from its origin within the corolla tube. The fully exposed seeds mature and dehisce after the placenta separates from the corolla tube, the parchment thin fruit carpels remain within the corolla tube. The non-shattering phenotypes we discovered disrupt this process. Placentas of 100% of the fruits of non-shattering individuals fail to rotate, so the seeds mature and dehisce within the corolla tube (Fig. 1). The dorsal abscission layer arises within these phenotypes, but most of the seed is retained within the corolla tube. The placenta dries over the developing seed, further decreasing shattering (Fig. 1). The carpels are torn irregularly by the expanding and maturing seed.

Since the selection of non-shattering progeny within VL-119, we have developed progeny from these selections with fruits which fail to split open. Numerous S1 individuals have been tested from a promising selection. The corolla tubes of certain individuals within this S1 family did not split open; the corolla tube encased the fruit carpels, placenta, and mature seed, thereby conferring complete seed retention. Naturally, these progeny are exactly what we have been striving to develop and what everyone has been hoping to find. Their development has been pivotal for Cuphea.

The precise biological and genetic underpinnings of the non-shattering trait are unknown. The non-shattering phenotypes came from wildtype genetic backgrounds of the progenitor species. They are not the consequence of induced mutations, rather they are the consequence of wide hybridization which led to genotypes which cannot arise and phenotypes which seemingly do not arise within either species alone. At the very least, the natural diversity within a given species of Cuphea is probably not sufficient to lead to non-shattering progeny purely by hybridization and selection alone. We have not observed phenotypic differences within any species sufficient for developing non-shattering germplasm. The phenotypic diversity within VL-119 significantly increased with repeated intermating. Ongoing recurrent selection work should perpetuate this trend for many years since we have kept the 'effective population size' very great for certain selection projects. Because C. viscosissima x C. lanceolata hybrids are fertile, breeding them is no different from breeding intraspecific populations, save for understanding and compensating for reduced recombination (Brandt and Knapp 1992). Many additional nonshattering phenotypes should arise from VL-119 and other extensively intermated interspecific populations. Naturally, we are aggressively developing non-shattering populations, lines, and cultivars for the eventual commercialization of Cuphea.

As a whole, seed dormancy is less severe of a problem than seed shattering for Cuphea; however, seed dormancy is severe within wild populations of C. viscosissima. Seed dormancy has been observed within C. lanceolata, but it is not severe within many populations, and non-dormant populations and lines of this species have been developed (Knapp 1990; Knapp and Tagliani 1990). We have developed a fully non-dormant inbred line, (LN-61ND)S5, of C. lanceolata; freshly harvested seed of this line germinates. It is the only fully non-dormant line which has been reported within the genus.

Until the spring of 1990, breeding work with C. viscosissima was impeded by severe seed dormancy. The initial aim of our interspecific breeding work was to introgress genes for non-dormancy from C. lanceolata to C. viscosissima, while retaining the autofertility of C. viscosissima (Table 1). In retrospect, this was not especially difficult to achieve. We originally selected within BC1S1 populations using C. viscosissima, an autogamous species, as the recurrent parent. We have not recovered autofertile progeny within F2 populations; however, they arise with great frequency within BC1S1 populations. Using C. viscosissima as the recurrent parent circumvents the negative consequences of the genetic load of the C. lanceolata genome. As most allogamous species, C. lanceolata manifests inbreeding depression and heterosis (Ali 1991; Knapp et al. 1991). Breeding work is underway to develop autofertile C. viscosissima x C. lanceolata inbred lines for F1 hybrids which retain a significant percentage of the C. lanceolata genome and exploit the heterosis between C. viscosissima and C. lanceolata and within C. lanceolata.

The development of autofertile non-dormant lines has significantly affected our work to induce seed retention mutants within these species. We are advancing our mutation breeding work, despite the discovery of non-shattering phenotypes, because of the merits and utility of having induced mutations affecting fruit morphology and dehiscence, e.g., mutations which eliminate the dorsal suture of the mature fruits could be very useful. Our original objective was to mutagenize C. viscosissima, but severe seed dormancy has prevented us from developing and screening large mutagenized populations of this species. The only consistent way to overcome the seed dormancy problem within this species is to excise embryos from whole seeds after-ripened for two to three months. This was done to develop ~2,000 M2 lines of C. viscosissima. Seventy-two mutations affecting seed oil fatty acid percentages have been isolated from these lines (Knapp and Tagliani 1990; Knapp 1992). Non-shattering mutants were not observed among this limited sample of lines. Many more M2 progeny must be screened to recover mutations affecting fruit morphology or dehiscence. The seed dormancy problem was solved by the development of autofertile non-dormant C. viscosissima x C. lanceolata lines (Table 1). We have since used these lines to develop M2 populations originating from at least a quarter of a million mutagenized M1 individuals. The screening of these populations for mutations affecting fruit morphology and dehiscence and other traits should be completed within the next two years.

Mutation breeding of Cuphea has been underway for several years (Hirsinger 1980; Hirsinger and Roebbelen 1980; Campbell 1987; Roebbelen and von Witzke 1989; Roebbelen 1991). This was motivated by the lack of natural diversity for seed shattering, but non-shattering mutants have not yet been isolated. Roebbelen and von Witzke (1989) reported a 'radial' flower mutant of C. calophylla, but this phenotype was not shown to affect fruit dehiscence or to decrease seed shattering (Roebbelen and von Witzke 1989). Seed shattering seems to be as great for the radial flower phenotype as for the wildtype of C. calophylla. A shift from zygomorphic to actinomorphic flowers should not be presumed to lead to seed retention.

Roebbelen and von Witzke (1989) screened M2 progeny from 9,312 M1 progeny of C. tolucana and M2 progeny from 3,693 M1 progeny of C. wrightii. These progeny were developed by EMS mutagenesis. Mutations affecting fruit dehiscence were not observed within either species; however, mutations affecting several morphological traits were observed within C. tolucana (Roebbelen and von Witzke 1989). Mutation rates for C. tolucana and C. wrightii were 1.4 and 0.0% (Roebbelen and von Witzke 1989). The lack of mutations within C. wrightii is not surprising. This species is undoubtedly an allotetraploid (2n = 4x = 44) and, as such, the phenotypes of most induced mutations are masked by duplicate genes. Additional evidence for this comes from the mutagenesis experiments of Campbell (1987) who observed no mutant phenotypes within C. wrightii using mutagen dose rates several fold greater than those used to induce mutations within the diploid species C. tolucana (2n=2x=20).

The question of inducing non-shattering mutants is still very much open. Extensive breeding work with an autogamous diploid species, e.g., C. viscosissima x C. lanceolata or C. viscosissima, is essential to maximize the probability of isolating non-shattering mutants, and this has not been done.


Many productive Cuphea species are allogamous, e.g., C. lanceolata (Knapp et al. 1991), C. laminuligera (Krueger and Knapp 1991), and C. leptopoda. Most allogamous species of Cuphea are pollinated by bumblebees and other Hymenopteran and Lepidopteran insects. The long length of the floral tubes of most allogamous species prevents honeybees from foraging for nectar. Ironically, C. viscosissima, an autofertile species with a fairly short floral tube (Graham 1988), is regularly pollinated by honeybees. Honeybees do forage within C. lanceolata, C. leptopoda, and many other large flowered allogamous species, but not effectively.

Insect-pollinated allogamous species of Cuphea probably cannot be produced commercially. Even if honeybees or some other domesticated pollinator effectively pollinated these species, their commercial use is impractical and prohibitively expensive. This was a major factor which led us to select C. viscosissima for intense domestication and breeding work.

Because it is feasible to transfer genes between C. viscosissima and C. lanceolata (Table 1), we initiated breeding work to exploit the hybrid vigor of C. lanceolata (Ali 1991) by developing autofertile inbred lines for F1 hybrids. Naturally, by introgressing genes for autofertility from C. viscosissima to C. lanceolata, the subsequent inbreeding within autofertile lines leads to inbreeding depression within those lines or populations where a significant percentage of the C. lanceolata genome is retained. These lines can be effectively exploited by using F1 hybrids; however, mechanisms have not yet been developed for producing F1 hybrid seed of Cuphea.


Most Cuphea species are characterized by sticky or glandular hairs covering their stems, leaves, and flowers (Graham 1988; Amarasinghe et al. 1991). These hairs have been repeatedly cited as a negative trait which must be eliminated to advance Cuphea (Hirsinger 1980; Hirsinger and Roebbelen 1980; Thompson 1984). The stickiness of Cuphea is unpleasant, but it is not a barrier to the commercialization of Cuphea. Although some of the sticky residue from Cuphea chaff accumulates in harvesting equipment, it does not seem to hinder harvesting; however, commercial scale tests have not yet been done. Sticky non-shattering Cuphea cultivars can be handled and harvested like other summer annuals, such as sunflower or soybean, by direct combining the crop after killing frosts destroy the foliage. Whether or not glabrous or non-sticky cultivars should be used is a subject for debate. Sticky hairs may be an effective defense against many insect pests. Aphids and many other insects are immobilized by the sticky hairs of C. viscosissima and C. lanceolata. At the very least, many insects cannot navigate wildtypes of many Cuphea species, among them C. viscosissima and C. lanceolata, and eliminating the sticky hairs might increase their vulnerability to many insect pests. Non-sticky mutants have been reported for C. lanceolata (Hirsinger 1980; Hirsinger and Roebbelen 1980), and the non-sticky trait might prove useful, but additional work is needed to determine the role of sticky hairs as a defense mechanism against insect pests before non-sticky cultivars are used. Regardless, stickiness should be repeatedly branded as a negative trait. It might ultimately be vindicated by serving as a defense against an otherwise serious insect pest. Indeed, it might be useful to breed for increased hairiness and stickiness, which is an unpleasant thought for anyone who has worked with this genus.

In addition to sticky hairs, indeterminate flowering has been cited as a negative trait of Cuphea (Hirsinger 1980; Hirsinger and Roebbelen 1980; Thompson 1984). It might be useful to develop determinate flowering Cuphea, but indeterminate flowering poses no problem for the production or harvest of Cuphea. Indeterminate flowering has not been shown to positively or negatively affect the seed yields of Cuphea. This cannot be determined until determinate flowering phenotypes are discovered or developed. Determinate flowering phenotypes might well be lower yielding than indeterminate flowering phenotypes.

Crop architecture has been cited as another problem trait (Roebbelen and von Witzke 1989); however, the wildtype crop architectures of C. viscosissima and C. lanceolata pose no problem for the production or harvest of Cuphea, nor do they negatively affect seed yield. These species grow upright, and very strong upright growth can be achieved by planting densely. It might be useful to modify the architecture of these species, e.g., to develop monoculm cultivars, but the merits of a 'monoculm' architecture (Roebbelen and von Witzke 1989) are uncertain. Furthermore, useful monoculm phenotypes have not yet been developed. Hirsinger (1980) and Roebbelen and von Witzke (1989) reported a monoculm mutant of C. lanceolata, but this mutant is unproductive and inferior to the wildtype. The negative characteristics of this phenotype seem to be a consequence of pleiotropy. Additional work to modify architecture might be useful, but a monoculm architecture is not necessarily going to lead to seed yield increases.


Estimates of coconut and palm kernel oil imports for North American and Europe from 1986 to 1988 (Mackie and Calhoun 1991) can be used to estimate the demand for MCT-rich oils (Table 2). North America and Europe imported 1.72 billion kg of coconut and palm kernel oil per year from 1986 to 1988. Assuming a seed yield of 2,500 kg/ha and an extracted oil percentage of 25% for Cuphea, 100% of the North American and European demand for MCT-rich oil could be met by producing 2.75 million ha of Cuphea. With a price of $0.65/kg for the seed, the projected revenue from the seed alone is 1.12 billion dollars. Cuphea cannot be expected to capture all of the market, but it should capture some of it. Furthermore, it might significantly restructure the market because of the many new kinds of oils it can supply (Knapp 1992). Although these demand estimates are rather crude, they demonstrate the underlying economics of MCTS.

The demand for MCT-rich oils is expected to steadily increase over the next several decades as the world population increases. As documented by Mackie and Calhoun (1991), world trade in oils from plants and oilseed meals "tripled from 1962 to 1988, while the number of countries participating in world trade increased by 85%." Mackie and Calhoun (1991) further stated, "Oilseed, oil, and meal consumption in most countries outpaced production during this period. Consequently, more countries have turned to world markets for a growing part of their oilseed product needs. These developments not only increased the volume of world trade in oil products between a larger group of countries, but also changed the pattern and direction of trade flows for these products." Cuphea could become a significant factor in oilseed trade over the next several years since virtually 100% of the MCT needs of many countries are being met by imports.

These demand estimates for MCT-rich oils do not account for new food and industrial uses of caprylic, capric, lauric, and myristic acid-rich oils from Cuphea. The great diversity of fatty acid phenotypes of Cuphea could greatly impact the demand for different oils (Knapp 1992). MCTs are presently used for infant feeding and hyperalimentation, especially for the critically ill, and they have been shown to decrease heart disease, breast cancer, colon cancer, and other diseases when used as the primary dietary lipid source (Babayan 1981; Bach and Babayan 1982; Babayan 1987). As a consequence, the use of MCTs in the human diet might increase if an inexpensive source such as Cuphea oil is developed. Present use of MCTs is severely restricted by the cost of synthesizing them from 8:0 and 10:0 fractions of coconut and palm kernel oils.


The development of autofertile non-dormant and non-shattering germplasm has created the basis for developing profitable cultivars of Cuphea. Although much effort remains in order to commercialize Cuphea, the development of non-shattering germplasm eliminates the risk which has heretofore made investments by agribusiness and state agricultural experiment stations risky and impractical. As the risk further diminishes over the next several years, cultivar testing and agronomic extension must be developed to give farmers the information needed to maximize seed yields and profitability, while processing and distribution networks must be developed. Beyond this, more plant breeders must endeavor to advance this crop. Decreased risk should catalyze an increase in the scale of breeding, genetic, and biotechnology efforts within this important and useful genus. Significant expansion cannot take place, however, until the commercial promise of autofertile non-shattering cultivars is demonstrated.


*This work was funded by grants from the USDA (58-5114-9-1002 and 91-37300-6569), Soap and Detergent Association, and Procter and Gamble Company. Oregon Agricultural Experiment Station Technical Paper 9994.
Table 1. History of Cuphea domestication and breeding work at Oregon State University marking the major events which led to domesticated germplasm of C. viscosissima x C. lanceolata f. silenoides.

Date Event
August of 1987 Fertile C. viscosissima x C. lanceolata progeny are discovered.
January of 1988 C. viscosissima and C. lanceolata are selected as targets for domestication. Breeding work within these species is significantly increased, while work with other species ceases.
March of 1990 Autofertile non-dormant phenotypes are discovered among C. viscosissima x C. lanceolata BC1S1 progeny.
September of 1991 Non-shattering phenotypes are discovered within the C. viscosissima x C. lanceolata population VL-119.
December of 1991 Fully non-dormant autofertile C. viscosissima x C. lanceolata lines are developed.
May of 1992 Non-splitting phenotypes are observed within S1 families from non-shattering individuals selected from VL-119.

Table 2. Mean coconut (copra) and palm kernel oil imports by North America and Europe from 1986 to 1988. Import estimates were compiled from USDA/ERS statistics (Mackie and Calhoun 1991).

Imports (1,000 Tonnes)
Importerz Commodity 1986 1987 1988 Mean
North America Palm Kernel Oil 181 196 233 203
Europe Palm Kernel Oil 351 354 395 367
North America Copra Oil 558 533 470 520
Europe Copra Oil 645 639 601 628
Total 1,735 1,722 1,699 1,718
zStatistics for North America are sums for the United States, the U.S. Virgin Islands, Canada, Puerto Rico, St. Pierre, and Miquelon. Statistics for Europe are sums for the EC10 (Belgium-Luxemborg, Denmark, France, the Federal Republic of Germany, Greece, Ireland, Italy, the Netherlands, and the United Kingdom), other western European countries (Andorra, Austria, the Faeroe Islands, Finland, Gibraltar, Greenland, Iceland, Malta, Norway, Portugal, Spain, Sweden, and Switzerland), and eastern European countries (Albania, Bulgaria, Czechoslovakia, the German Democratic Republic, Hungary, Poland, Romania, USSR, and Yugoslavia) (Mackie and Calhoun 1991).

Fig. 1. Non-shattering phenotypes from the C. viscosissima x C. lanceolata f. silenoides population VL-119.

Fig. 2. History and pedigree of the C. viscosissima x C. lanceolata f. silenoides population VL-119.

Last update September 12, 1997 aw