Index | Search | Home | Table of Contents

Hof, L. 1996. Dimorphotheca pluvialis: A new source of hydroxy fatty acid. p. 372-377. In: J. Janick (ed.), Progress in new crops. ASHS Press, Arlington, VA.

Dimorphotheca pluvialis: A New Source of Hydroxy Fatty Acid

Lysbeth Hof

    1. Synchronization Between Plants
    2. Synchronization Within Plants
    3. Oil Content
    4. Pollen Transfer
  4. Table 1
  5. Fig. 1
  6. Fig. 2
  7. Fig. 3

In the search for alternative crops for Dutch agriculture, Dimorphotheca pluvialis L. (Mnch), Asteraceae, is being considered as a potential new crop with industrial applications. Its seeds contain oil with 60% to 65% dimorphecolic acid (d9-hydroxy,t10,t12-octadecadienoic acid): a hydroxy fatty acid with two conjugated double bonds. This feature provides dimorphecolic acid with a unique functionality and properties that are totally different from other known hydroxy fatty acids as ricinoleic and lesquerolic acid. The chemical structure suggests that the molecule should be very reactive, and hence suitable for a wide range of industrial products such as surface coatings, surfactants, plastic foams, or as additive in plastics (Knowles et al. 1965; Muuse et al. 1992). New applications and markets need to be developed, possibly leading to new products with a high added value.

Dimorphotheca pluvialis is a herbaceous annual native to South West Africa (Norlindh 1977). As is common in the Asteraceae, the capitulum (flower head) bears two types of florets. The species Dimorphotheca is characterized by hermaphrodite disc florets and female-fertile (male-sterile) ray florets. Both types of florets produce distinctly different types of seeds (achenes). Seeds produced by ray florets are small, angular, while those of the disc florets are flattened and have winged margins (Barclay and Earle 1965). The ray florets have one large white petal, which is often colored purple at the base, giving the appearance of a "ring" in the inflorescence.

The species is well adapted to the maritime climate of Northern and Western Europe, and fits in a rotation system with annual crops. Although it is known as a garden ornamental, Dimorphotheca is considered an undomesticated species showing many primitive characteristics. Populations collected from the natural habitat in general have a long, unsynchronized flowering period and show poor seed retention. These factors together account for severe yield losses prior to and during harvest. Realized yields at trial fields range 500-1500 kg/ha, with potential yields of at least 2000-2500 kg/ha. As it is sensitive to frost, in the Netherlands, Dimorphotheca is grown as a summer annual. It is sown in April, flowers in July and can be harvested in August (van Dijk et al. 1993). The average oil content of collected populations is 21%, which is too low for mechanical expelling. At present, oil recovery should be done by solvent extraction or preferably with supercritical carbon dioxide extraction (Muuse et al. 1992).

Dimorphotheca was first introduced in the Netherlands by the Dutch Gene Bank (CGN) in 1986, and since 1990 has been studied extensively in the framework of three large multidisciplinary projects (the Dutch National Oilseeds Program, and the EC-projects VOICI and VOSFA). In these projects expertise from the whole production chain was brought together including germplasm collection, evaluation, cultivation, breeding, crop physiology, pathology, harvest techniques, oil recovery, processing of the oil, application, and market research (van Soest and Mulder 1993).

At the Centre for Plant Breeding and Reproduction Research (CPRO-DLO), research in Dimorphotheca has been focusing on improvement of synchronization of flowering and seed ripening, oil content, and plant architecture. Furthermore, research is carried out to study pollination and mating system.


In order to determine optimal selection strategies for synchronization of flowering and oil content, variation and heritability of these characters were estimated. Flowering synchronization was considered particularly important. Large differences in time of flowering make it difficult to detect slight differences in seed retention, hence making selection for this character at present almost impossible. More synchronized populations are expected to have a shorter flowering and seed ripening period, facilitating the determination of the optimal harvest date. A too early harvest causes reduction of seed yield because of a large proportion of immature seeds. If harvest is carried out too late, yield is severely reduced because of seed losses due to shattering.

For synchronization of flowering two distinct characteristics were distinguished:

  1. The synchronization between plants, which was defined as the period of time between sowing and the first flower to open. A population flowers synchronously if all plants start flowering at the same time.
  2. The synchronization within plants, which was defined as the period of time in which individual plants produced 90% of their total number of flowers. A plant flowers synchronously if it produces its flowers in a relatively short period of time.
Both the synchronization between plants of a population, as well as the synchronization within plants of that population are considered to have a large effect on the synchronization of flowering of a population.

To establish the variation for synchronization of flowering (both between and within plants) and oil content, two experiments were carried out on loam soil at location Lelystad in the Netherlands. Heritabilities were estimated by means of parent-offspring regression in the following year.

Synchronization Between Plants

In 1992, 350 plants of population 883168 were sown at a wide density (50 x 50 cm), allowing scoring and harvesting of individual plants. Time of flowering was scored every Monday and Thursday, and for further analyses expressed as the number of days from sowing until first open flower. The mean time of flowering was 78 days after sowing (range: 68-90 days). Although most plants started to flower after 75 to 80 days, some started to flower one week earlier, others as much as three weeks later. This means there was a difference of four weeks between the first and last plant to open its first flower.

To estimate the heritability of this character in this population, 40 plants were selected and their seeds collected. D. pluvialis is considered a predominantly outcrossing species, and the progenies of the selected plants are considered to be half-sib families. In 1993, a trial field was sown with 24 (three rows of eight) plants of each of the 40 families, and time of flowering was scored.

The relationship between selected plants and the mean of their progenies for time of flowering is presented in Fig. 1. It is clear that selected plants and their corresponding families showed considerable resemblance, this despite difference in weather conditions in 1992 and 1993.

Narrow sense heritability (h 2n) is described as the ratio between genotypic and phenotypic effects. High heritabilities indicate that the genetic component in the observed phenotype is relatively important to the environmental component. This means that high heritabilities for a character usually result in a quick response to selection. With low heritabilities the response to selection is not necessarily lower, but more time consuming. From the linear regression of offspring (Y) on female parents (X), expressed as Y = a + bX, the (narrow sense) heritability can be estimated by h 2n = 2b (Falconer 1989). The estimated heritability for time of flowering in this experiment was 0.94, which is very high.

Synchronization Within Plants

For estimation of duration of flowering of individual plants a similar experimental lay out was used. In 1992, a field was sown with 220 plants of an unselected population (879585). Plant spacing was approx. 100 x 100 cm. Twice a week the number of open flowers per plant (NOF) was scored. At the end of the growing season plants were removed from the field and the total number of heads per plant (TNH) was counted. Knowing the total number of heads per plant, a correction could be made for the error caused by counting open flowers twice on two consecutive counting dates. In general, flowers stay open about 4 to 6 days. For each plant a correction term (CT) was estimated, being the ratio between the actual total number of heads (TNH) and the mathematical sum of counted open flowers on the counting dates (MNF). This ratio is 1 if no flowers are counted twice. In this experiment, in most plants the ratio was estimated between 1.0 and 2.0, meaning that 0% to 100% of the flowers were indeed counted twice on two consecutive counting dates. The number of newly opened flowers per counting date (NNOF) was estimated by the product of the number of counted open flowers (NOF) and the estimated correction term (CT): NNOF = NOF x CT, where CT is TNH/MNF.

The cumulative numbers of open flowers per plant plotted against time fitted a logistic curve (Y = c/[1 + e-b(X - m)]). In this curve, c represents the upper asymptote (being the total number of heads, TNH), b the "slope parameter," and m the inflexion point of the curve, which is also the date at which the maximum number of open flowers was counted: peak bloom. The flowering of each individual plant was characterized by these three parameters. In this experiment this model on average accounted for 99.6% of the observed variation, indicating that it described the flowering of individual plants well.

Using this model, the period in which the plants produced 90% of their total number of flowers (the duration of flowering) could be calculated. The 90% interval, and not 100%, was chosen because slight deviations from the model occurred at beginning and end of flowering, accounting for relatively large aberrations in estimates of duration of flowering when using the 100% interval.

Duration of flowering ranged from 11 to 63 days, with a mean of 27 days. Since plants did not start flowering at the same time, environmental factors may have had a considerable effect. Therefore, also for this character heritability was estimated by means of parent-offspring regression.

From the population grown in 1992, 20 plants were selected showing much variation for duration of flowering. In 1993, ten plants per progeny were sown in a complete randomized block design, and flower counts were made in the same way as before. The explained variation for the fit of the logistic model on the data was 99.8%.

The relationship between female parents and the mean of their corresponding families is shown in Fig. 2. The calculated regression line explained only 13% of the variation. This means that it leaves 87% of the variation still to be accounted for, and seems drawn rather arbitrarily through a cloud of data points. The estimated heritability (based on the slope of this regression line) of 0.27 can therefore be regarded as unreliable.

Oil Content

The same trial field as described in the first experiment was used to assess the variation and heritability for oil content. Of all plants seeds were harvested and separated in winged and unwinged seeds. Oil content of the winged seeds was measured with Near InfraRed Spectroscopy (NIRS) equipment (InfraAlyzer 500, Bran+Lübbe). Oil content of individual plants ranged from 15% to 29%, with a mean of 21.5%. From this experiment 40 plants were selected representing almost the whole range. Progenies were tested in 1993 (same experimental lay out as in time of flowering experiment). Seeds were collected and oil content was measured with NIRS. The relationship between the female parents and the mean of the corresponding offspring is presented in Fig. 3. The estimated heritability was 0.36 but the regression line explained only 17% of the observed variation.

Pollen Transfer

Dimorphotheca pluvialis is considered a predominantly outcrossing species, however little is known on the mode of pollen transfer. The influence of insects on seed set was investigated in a complete randomized block design with three replications, three populations, and three treatments. The treatments consisted of: (1) plots in open air (free insect visitation), (2) plots with cages open at the North side (free insect visitation + shading effect of cage), and (3) plots with cages (no insects, shading effect). The second treatment was included as a control, to determine possible effects of the shading caused by the cages on growth and development of the crop. Plot size was 3 x 3 m. The number of open flowers was counted weekly on two subplots of 0.25 m2. Seed set was determined by picking 20 flowers randomly, and counting the number of winged and unwinged seeds.

Analyses of variance revealed no difference between treatments 1 and 2 for crop development, seed set and thousand seed weight. Apparently the light shading did not effect these characters (Table 1). Population 879585 flowered slightly earlier than the other two. No population x treatment interaction was found for any of these characters.

Seed yield of plots with open cages was lower than yield of open fields. This could not be explained by a lower seed set, lower number of flowers, or lower thousand seed weight. In the harvest bags of this treatment moths were found, which might have caused severe damage.

Exclusion of insects led to a prolonged flowering of the crop, and a severely reduced seed set and yield. Thousand seed weight was higher.


For time of flowering, duration of flowering, and oil content of seeds, sufficient variation was found to enable improvement by means of selection. Heritability for time of flowering appears to be high, indicating that selection for this character will show quick response. For duration of flowering and oil content of the seeds, heritability estimates by means of parent-offspring regression were questionable, but most likely these heritabilities are not very high. In this case the result of selection does not necessarily have to be less, but is more time consuming. The environmental component of the observed phenotype is relatively large, and therefore may conceal the genotypic component.

For these experiments it was assumed that random mating and complete cross pollination has taken place. Furthermore, interaction effects (epistasis, genotype-year, year-location, and genotype-location) were considered negligible. It is likely that some of these assumptions were incorrect, and may have affected the outcome. Year and location effects can only be studied when experiments are carried out at several locations in several years. The presented results on heritability estimates are therefore preliminary, but nevertheless give an indication of what can be expected from selection.

Presence of insects during flowering is essential for a good seed set. Exclusion of insects may result in yield losses up to 75%.

Dimorphotheca pluvialis is as yet not ready for commercialization. Several agronomic constraints are recognized, but most can be overcome given time. Other problems still lay in the area of oil recovery and purification. However, the unique structure of dimorphecolic acid justifies further studies.


Table 1. Effect of exclusion of insects during flowering on crop development and yield characteristics of Dimorphotheca pluvialis.

No. of open flowers/
0.25 m2
Thousand seed weight
Variable June 30 July 7 July 14 July 21 July 28 Seed yield (g/m2) No. unwinged seeds/flower No. winged seeds/flower unwinged seeds (g) winged seeds (g)
879127 1.2 28.3 140.9 126.0 22.8 70.7 10.7 29.0 3.39 2.26
879731 0.7 29.6 134.6 107.0 26.1 85.7 10.3 30.2 3.50 2.28
879585 2.2 48.3 195.3 74.1 30.9 79.8 11.1 28.4 3.48 2.25
Sign.z ** *** *** *** x * NS NS NS NS
Cage treatmenty
1 1.4 40.6 148.9 82.9 7.5 125.9 14.4 41.3 2.90 2.14
2 0.9 33.8 142.1 71.6 4.0 83.0 15.4 41.2 3.03 2.20
3 1.7 31.9 179.7 152.6 68.4 27.4 2.4 5.1 4.46 2.46
Sign.z NS NS * *** x *** *** *** *** ***
zNS = effect not statistically significant, ***significant 0.1%, ** at 1%, * at 5%.
yTreatments: 1 = open plots (insects yes, shading no), 2 = partly opened cages (insects yes, shading yes), 3 = closed cages (insects no, shading yes).
xDistribution of residuals not Normal, ANOVA performed on Ö-transformed data.

Fig. 1. Parent-offspring relationship for beginning of flowering in a population of Dimorphotheca pluvialis (days after sowing).

Fig. 2. Parent-offspring relationship for duration of flowering in a population of Dimorphotheca pluvialis (days).

Fig. 3. Parent-offspring relationship for oil content in a population of Dimorphotheca pluvialis (% oil in the seed).

Last update August 21, 1997 aw