Towards domestication of Dimorphotheca pluvialis : studies on the genetic improvement of a potential oilseed crop for industrial applications

World-wide, but particularly in Western Europe and the USA, the interest in arable crops for non-food use has increased substantially over the past few decades. Surpluses of the major food crops and the industrial interest for renewable resources have led to research and development programmes aiming at the introduction of crops with industrial applications. Particularly vegetable oils with fatty acids containing functional groups are very attractive as substitutes for mineral oils used in the production of e.g. lubricants, surfactants, coatings and polymers. Therefore, in recent years much effort has been made to domesticate wild species containing such oils.

Although the domestication and development of each new oilseed crop requires a specific approach dependent on the characteristics of the plant species, the history of the crop and its potential uses, some generalisations can be made about the steps involved. Both for crop development (agricultural side) and product development (industrial side), four stages can be distinguished: exploration, examination, expansion and exploitation. A more detailed description of this concept, known as the '4-ex model', can be found in Chapter 1 of this thesis.

In 1986, the first of a series of projects on potential industrial crops in the Netherlands was launched. Around 40 oilseed species were evaluated for various agronomic characteristics as well as oil content and quality. After considering the industrial interest and agricultural potential, one of the species selected for further breeding research was Dimorphotheca pluvialis (L.) Moench. Seeds of this species contain ca 21% oil with approximately 60-65% dimorphecolic acid (Δ9-hydroxy,10t,12t-octadecadienoid acid). The highly reactive hydroxydiene structure provides this fatty acid with a unique functionality and properties, making it potentially suitable for application in e.g. pharmaceuticals, surfactants, coatings, plastic foams, polymers, fragrances and flavours.

Although species of the genus Dimorphotheca have been known in the Netherlands as garden ornamentals for several centuries, the use of D. pluvialis for the production of seed oil is completely new. With regard to its use as an arable crop no breeding activities have been reported earlier, and as such the species should be considered as undomesticated. In the exploration of its potential as an arable oilseed crop, several constraints were identified, e.g. a long and unsynchronised period of flowering and seed ripening, poor seed retention and rather low oil content of the seeds. Hence, reported seed and oil yields were erratic. Seed retention seemed difficult to quantify, and preliminary observations revealed little variation for this character. Given the relatively short duration of the projects on which this thesis is based, attention was therefore focused on other important yield limiting factors: flowering synchronisation and oil content. Emphasis was laid on genetic improvement of these traits by means of selection. The efficiency of selection was examined by determination of the response to selection and estimations of the heritability.

For successful seed production, knowledge on the mode of reproduction of the crop is imperative. In the literature, D. pluvialis is described as a highly allogamous species, but little is known about the mode of pollen transfer. The influence of insects on several yield components was studied by comparing plant populations in the presence and absence of insects (Chapter 2). Exclusion of insects had a dramatic effect on the production of flowers and the duration of the flowering period. The total number of flowers at peak bloom was higher, and flowering continued longer in the absence of insects. Seed weight was somewhat higher, but seed set, seed yield and oil content were severely reduced under these circumstances, thousand seed weight was somewhat higher. The total seed and oil yield of insect-visited plots were, respectively, 4.5 and 5.5 times higher than those of insect-free plots. These results confirm the assumptions on the allogamous nature of the species and underline the importance of insect pollination for adequate seed yield in D. pluvialis .

Next to seed yield, oil content is a second important oil yield determining factor. High and stable oil yields of good quality are essential to provide industry with a constant supply. Three different populations of D. pluvialis were therefore subjected to mass selection for higher oil content (Chapter 3). After three cycles of selection at an intensity of 10%, for all three populations a significant increase in oil content was observed. Per selection cycle, an average gain in oil content ranging from 0.5% to 1.2% was achieved, depending on the population used. Realised heritabilities for this feature after three selection cycles in these populations ranged from 0.15 to 0.58. In Chapter 4, heritabilities for oil content were estimated from parent-offspring regression and half-sib family variance components. For this, forty plants were selected and progenies were tested twice, in two consecutive years. Heritability estimates from this experiment were moderately low: 0.34 from parent-offspring regression and 0.27 from variance components. If from this experiment 10% of the parental plants had been selected, an increase in oil content of 0.8% in one selection cycle would have been achieved. These values are in accordance with the values mentioned in Chapter 3. Although heritability estimates are specific for populations and environmental circumstances, it is likely that (mass) selection for increased oil content in general will be effective, particularly in the early generations. Considering the observed additive genetic variation in the populations, an average oil content of at least 30% seems feasible.

Particularly in combination with poor seed retention, the long, unsynchronised period of flowering and seed ripening is undesired. When a crop is harvested too early, yield losses occur due to incomplete setting and maturation of the seeds. When harvested too late, however, seed shattering will account for a severe reduction of seed yield. With regard to synchronisation of flowering, two main components can be distinguished; i.e. the synchronisation between plants and the synchronisation within plants. Synchronisation between plants is attained when plants of a population start flowering at the same time. Synchronisation within a plant is achieved when its flowers are produced in a short period of time. Both components are considered important for improvement of flowering synchronisation of the crop. Therefore, the flowering of individual plants was studied by counting the open flowers at regular time intervals (Chapter 5). It appeared that the flowering process of individual plants can be described mathematically by a logistic curve, obtained by the regression of the cumulative number of open flowers plotted against time. The curve is characterised by three parameters, corresponding with the total number of flowers produced by the plant, the rate of flowering development and the day at which peak bloom was reached. From these parameters, subsequently two other characteristics could be derived, namely onset of flowering and duration of flowering within the plant. Similar to the method described for oil content, heritabilities of the flowering traits were estimated by using parent-offspring regression and half-sib family variance components analyses. Onset of flowering and peak bloom showed high (> 0.69) heritabilities for both methods and both years, and total number of flowers showed moderate to high (0.30 - 0.90) heritability values. For these traits considerable progress may be expected from mass selection, particularly in the early selection generations. Duration of flowering showed low to moderate values (0.25 - 0.45), and thus for improvement of this trait methods other than mass selection should be considered.

Duration of flowering, total number of flowers and onset of flowering do not seem to be correlated (Chapter 5). Oil content and onset of flowering also seem to be uncorrelated (Chapter 4). Selection for either of these traits will most likely not influence the others.

Modern agriculture requires uniform plant populations. The currently available populations of D. pluvialis , however, most often originate from botanical gardens or gene banks and show considerable variation for morphological and agronomic traits. To improve morphological uniformity and to determine a preliminary ideotype for plant architecture, divergent mass selection for this character was carried out (Chapter 6). In order to minimise undesired side-effects due to assortative mating caused by variation in onset of flowering, selection for plant architecture was combined with selection for onset of flowering (earliness). Hence, six selection groups were distinguished: all possible combinations of two plant architecture types (erect and procumbent), with three earliness classes (early, middle and late). Three cycles of combined selection resulted in a significant response for both traits in both directions, even at a low selection pressure. In this experiment, selection for early flowering or procumbent architecture showed a better response and a higher heritability than selection for late flowering or erect plant architecture. The different plant architecture selections showed similar flowering development and seed yield. Therefore, from these results no conclusions on ideal architecture type with regard to breeding for increased yield could be drawn. However, for cultivation generally erect plant types are preferred. Earliness did have a significant effect on seed yield: early flowering types showed the highest yields. As the yield experiment was carried out only in one year, and genotype by environment interactions could not be assessed, no firm conclusions on ideotype with regard to earliness could be drawn. Nevertheless, (very) late flowering selections in general are undesired in the Netherlands because of an increased risk of experiencing unfavourable weather conditions during flowering and seed set.

D. pluvialis seems well adapted to the climatic conditions of north-west Europe, and fits well in a crop rotation system with annuals (Chapter 7). Its susceptibility to soil-borne diseases should be taken into consideration, but so far this has not caused major crop damage. Other pests and diseases seem to be easily controlled by agrochemicals. Improvement of resistance to several diseases can most likely be achieved by breeding, and deserves further attention. Genotypes with quick soil cover and improved harvest index may contribute to a higher potential seed production. Apart from further selection for increased oil content and flowering synchronisation, special attention should be directed towards improvement of seed retention. Even under optimal harvest conditions, seed losses of 20% were reported, entirely due to shattering. For good oil quality, relatively expensive methods for oil recovery have to be used. At present, oil of D. pluvialis seems particularly suitable for use in products with a high added value. However, many potential applications have not been explored yet. The unique structure and functionalities of dimorphecolic acid call for further research!