Agronomic and seed quality evaluation of Camelina sativa in western Canada

Transcription

1 Agronomic and seed quality evaluation of Camelina sativa in western Canada R. K. Gugel and K. C. Falk Agriculture and Agri-Food Canada, Saskatoon Research Centre, 107 Science Place, Saskatoon, Saskatchewan, Canada S7N 0X2 ( Saskatoon Research Centre Contribution No Received 13 May 2004, accepted 5 June Gugel, R. K. and Falk, K. C Agronomic and seed quality evaluation of Camelina sativa in western Canada. Can. J. Plant Sci. 86: Renewed interest in Camelina sativa is primarily due to the unique fatty acid profile of the seed oil and its potential value in industry, cosmetics and human nutrition. To exploit C. sativa in western Canada, more information is needed on the performance of this crop in this region. Following a preliminary evaluation in 2001, replicated agronomic trials were conducted in 2002 and 2005 with 19 C. sativa and three oilseed Brassica accessions at Saskatoon and Scott, Saskatchewan and Beaverlodge, Alberta. The C. sativa accessions matured relatively early and were more tolerant of drought and flea beetle infestations than the Brassica oilseeds. Some C. sativa accessions had seed yields competitive with those of the Brassica oilseeds, but seed size was significantly smaller. Seed yields and oil contents of all crop species tested were highest at Beaverlodge, the most northern location. The Brassica oilseeds generally had higher oil contents than C. sativa; the highest oil contents of each crop species tested were associated with the lowest protein contents. In general, average oil and protein contents for C. sativa ranged from 38 to 43% and from 27 to 32%, respectively; for the Brassica checks, oil and protein contents ranged from 38 to 53% and from 21 to 33%, respectively, across all species. Variation in fatty acid composition was higher among the C. sativa accessions than among locations, but overall the ranges of individual fatty acids were relatively narrow. The most abundant fatty acids were oleic ( %), linoleic ( %), linolenic ( %) and eicosenoic ( %). The prospects of developing improved C. sativa germplasm for particular western Canadian environments are good; of primary importance are increased seed size and oil content. Additionally, stand establishment, fertility requirements and broadleaf weed control options need to be investigated. Acceptance of this species as a new oilseed crop for western Canada will also require developing sustainable markets for the oil and meal. Key words: Camelina sativa, seed quality, agronomic trait, oil and protein content, fatty acid Gugel, R. K. et Falk, K. C Évaluation de la valeur agronomique et grainière de Camelina sativa dans l ouest du Canada. Can. J. Plant Sci. 86: Si Camelina sativa suscite aujourd hui un regain d intérêt, on le doit essentiellement à la composition en acides gras unique de l huile extraite de sa graine et à son utilité éventuelle pour l industrie, le secteur des cosmétiques et la nutrition humaine. Avant d entreprendre l exploitation de cette culture dans l ouest du Canada, on a cependant besoin d en savoir plus sur son rendement localement. Après une évaluation préliminaire en 2001, les auteurs ont procédé à des essais agronomiques sur 19 obtentions de C. sativa et trois oléagineux du genre Brassica à Saskatoon et à Scott, en Saskatchewan, ainsi qu à Beaverlodge, en Alberta en 2002 et Les obtentions de C. sativa parviennent à maturité relativement tôt et tolèrent mieux la sécheresse et les infestations d altise avec répétitions des crucifères que les oléagineux du genre Brassica. Quelques obtentions de C. sativa ont donné un rendement grainier voisin de celui des oléagineux du genre Brassica, mais leurs semences étaient significativement plus petites. Toutes les espèces testées ont donné leur meilleur rendement grainier et la plus forte teneur en huile à Beavelodge, le site le plus au nord. Les oléagineux du genre Brassica produisent généralement plus d huile que C. sativa; pour chaque espèce testée, la plus forte concentration en huile est associée à la concentration de protéines la plus faible. Dans l ensemble, la teneur moyenne en huile et en protéines de C. sativa varie respectivement de 38 à 43 % et de 27 à 32 %; la teneur en huile et en protéines des témoins du genre Brassica s établissait entre 38 et 53 % et 21 et 33 %, respectivement, pour l ensemble des espèces. La composition des acides gras varie plus parmi les obtentions de C. sativa que d un endroit à l autre, mais dans l ensemble, la fourchette de valeurs des divers acides gras est relativement étroite. Les acides gras les plus abondants sont les acides oléique (12,8 à 14,7 %), linoléique (16,3 à 17,2 %), linolénique (36,2 à 39,4 %) et eicosénoïque (14,0 à 15,5 %). Le développement de meilleures variétés de C. sativa pour certains milieux de l Ouest canadien laisse entrevoir de bonnes perspectives; on devra surtout s efforcer d accroître le calibre et la teneur en huile des semences. L établissement des peuplements, les besoins de fertilisation et la lutte contre les mauvaises herbes dicolylédones sont des aspects qu il faudra examiner. Enfin, on devra créer des débouchés durables pour l huile et le tourteau de cette espèce si l on veut qu elle ait sa place parmi les oléagineux cultivés dans l ouest du Canada. Mots clés: Camelina sativa, qualité des semences, caractères agronomiques, teneur en huile et en protéines, acide gras Archeological evidence suggests that cultivation of Camelina spp. (Brassicaceae) as an oilseed crop began in southeast Europe in the late Neolithic (Knörzer 1978) and became well established in this region during the Bronze Age (Bouby 1998). By the Iron Age, camelina was a common crop across much of Europe and Scandinavia (Knörzer 1978; Bouby 1998); Camelina sativa (L.) Crantz was probably developed from these early plantings (Knörzer 1978). The importance of camelina as a food crop declined during the Middle Ages (Knörzer 1978), but cultivation continued 1047

2 1048 CANADIAN JOURNAL OF PLANT SCIENCE sporadically until modern times. More recently, C. sativa was cultivated on a relatively small scale in Europe and Russia until the mid-20th century (Zubr 1997). Camelina sativa was probably introduced to the Americas as a weed in flax (Putnam et al. 1993). European immigrants to the New World would have brought knowledge of C. sativa with them, but the species does not appear to have been cultivated except, perhaps, in small isolated plantings. In the 19th century Porcher (1863) described the favourable agronomics of C. sativa for the central and southern United States including low inputs, high yield, frost and drought tolerance and resistance to insect pests. A broad range of nutritional, industrial and medicinal uses also were described for the oil, oil cake and chaff, but despite the recommendation to cultivate C. sativa (Porcher 1863) little, if any, was grown. A 30-yr research program initiated in Minnesota in the mid-20th century added significantly to the body of knowledge on cultivation and use of C. sativa in the United States (Robinson 1987). Trials conducted in Canada from 1958 to 1960 (Plessers et al. 1962) produced results similar to those obtained in the United States and led to the suggestion that camelina be given serious consideration as a potential oilseed crop for northern latitudes. However, in spite of all the favourable reports, a market for the seed was lacking (Robinson 1987). Renewed interest in C. sativa has surfaced in Europe (Zubr 1997; Hebard 1998; Leonard 1998; Bonjean and Le Goffic 1999), North America (Putnam et al. 1993) and Australia (Francis and Campbell 2003) primarily due to the unique fatty acid profile of the seed oil and its many potential applications in industry, cosmetics and human nutrition. The fatty acid composition is largely unsaturated (> 90%) with significant amounts (30 40%) of linolenic acid (C18:3) and relatively low amounts (3 4%) of erucic acid (C22:1) (Budin et al. 1995; Zubr and Matthäus 2002). Furthermore, the seed meal is relatively low in glucosinolates (13 36 µmol g 1 dry seed) compared with other crucifers (Schuster and Friedt 1998) and has a favourable balance of amino acids (Zubr 1997; Bonjean and Le Goffic 1999), making the meal a potentially valuable feed for poultry, swine and ruminants (Putnam et al. 1993; Zubr 1997; Schuster and Friedt 1998; Bonjean and Le Goffic 1999). Camelina sativa is adapted to a wide range of climatic conditions, as demonstrated by its long history of cultivation across the European continent and its more recent favourable performance in North American and Australian evaluation trials. The multitude of potential uses for the crop, combined with its favourable agronomics, suggests that cultivation of C. sativa may have a place in western Canadian agriculture especially now that growers are being encouraged to diversify their cropping strategies. However, few agronomic evaluations of C. sativa have been conducted in western Canada and little published information is available for this region. Plessers et al. (1962) and Paul et al. (2000) reported a limited amount of yield and seed quality data from two locations in Alberta; however, significant effects of environment on seed yield (Vollmann et al. 1996; Francis and Campbell 2003) and quality (Zubr and Matthäus 2002; Zubr 2003) make it inappropriate to conclude that the performance of C. sativa would be similar at other locations in western Canada. Additional variation in agronomic and seed quality traits also may be discovered by evaluating more accessions in different environments, thereby providing valuable information for breeding programs. The objective of the present study was to evaluate the performance of 19 C. sativa accessions at three locations in western Canada that historically have had different environments based on data from crop variety registration trials. The primary focus of the study was on crop maturity, seed yield, oil and protein content and fatty acid composition of the seed oil. MATERIALS AND METHODS Accessions Sources of seed for 19 accessions of C. sativa and one accession each of Brassica rapa L., B. juncea (L.) Czern. and B. napus L. are listed in Table 1. Accessions were obtained from Agriculture and Agri-Food Canada, Saskatoon Research Centre, Saskatoon, Saskatchewan, Canada, the Institut für Pflanzengenetik und Kulturpflanzenforschung, Genbank-Aussenstelle Malchow/ Poel, Germany and Plant Gene Resources of Canada, Saskatoon, Saskatchewan, Canada. The three Brassica accessions are commercial cultivars of Polish canola (B. rapa AC Parkland ), Argentine canola (B. napus AC Excel ) and condiment Oriental mustard (B. juncea AC Vulcan ) that were included in the study as benchmarks for agronomic performance and seed quality. Camelina sativa seed used for the study was from isolation tent increases conducted in the field at Saskatoon in 1999; certified commercial seed of the Brassica cultivars was used for the study. Field Trials A preliminary field trial was conducted in 2001 to determine whether planting and harvesting procedures routinely used for canola and mustard trials were suitable for C. sativa. The performance of four accessions of C. sativa (CS01 CS04) was evaluated in comparison to three Brassica cultivars (BR01, BJ01, BN01) at a research farm managed by Agriculture and Agri-Food Canada at Saskatoon, Saskatchewan, Canada. Prior to sowing the trial, the site was treated with the pre-emergence herbicide ethalfluralin (5% granular, 20.0 kg ha 1 fall applied); soil fertility also was tested in early spring and optimized for canola production. The experimental design was a randomized complete block with four replicate plots of each accession. Each plot was 6 m long and consisted of four rows spaced approximately 18 cm apart; the seeding rate was 100 seeds per 3 m of row and seeding depth was 2 to 3 cm. Each plot was flanked by two rows of barley to prevent intermingling of adjacent plots. Terbufos (5% granular, 8.0 kg ha 1 ) was drilled with the seed to control flea beetles (Phyllotreta spp.) during crop emergence; post-emergence infestations were controlled with deltamethrin (50 g L 1 emulsifiable concentrate, 0.15 L ha 1 ) as required. All plots were trimmed to a uniform length prior to harvest. Plots were left standing in the field until they were completely ripe and were combine harvested without swathing; 4.34 m 2 of each plot was harvested.

3 GUGEL AND FALK CAMELINA SATIVA IN WESTERN CANADA 1049 Table 1. Accession number, country of origin and source of seed for Camelina sativa, Brassica rapa, B. juncea and B. napus accessions studied Accession Country of origin Source z C. sativa CS01 Former USSR AAFC-SRC, SRS 933 CS02 Former USSR AAFC-SRC, SRS 934 CS03 Poland IPK, CR 333 [CN ] CS04 Former Yugoslavia IPK, CR 382 [CN ] CS05 IPK, CR 464 [CN ] CS06 Former Yugoslavia IPK, CR 476 [CN ] CS07 IPK, CR 1671 [CN ] CS08 Germany IPK, CR 1674 [CN ] CS09 Germany IPK, CR 1675 [CN ] CS10 Denmark IPK, CR 1676 [CN ] CS11 IPK, CR 1678 [CN ] CS12 England AAFC-SRC, SRS 2989 [CN ] CS13 Poland AAFC-SRC, SRS 2380 [CN ] CS14 Former USSR PGRC, PGR 1186 [CN 30475] CS15 Former USSR PGRC, PGR 1187 [CN 30476] CS16 Former USSR PGRC, PGR 1188 [CN 30477] CS17 Former USSR PGRC, PGR 1189 [CN 30478] CS18 Former USSR PGRC, PGR 1190 [CN 30479] CS19 Canada PGRC, PGR [CN 45816] B. rapa BR01 Canada AAFC-SRC, AC Parkland [CN ] B. juncea BJ01 Canada AAFC-SRC, AC Vulcan [CN ] B. napus BN01 Canada AAFC-SRC, AC Excel [CN ] z AAFC-SRC = Agriculture and Agri-Food Canada, Saskatoon Research Centre, Saskatoon, Saskatchewan, Canada; IPK = Institut für Pflanzengenetik und Kulturpflanzenforschung, Genbank-Aussenstelle Malchow/Poel, Germany; PGRC = Plant Gene Resources of Canada, Saskatoon, Saskatchewan, Canada. [CN number] refers to the most recent accession coding from PGRC. Testing was expanded in 2002 and 2005 to include 19 accessions of C. sativa (CS01 CS19) and the same three Brassica cultivars that were grown in the 2001 trial. Field trials were conducted at Agriculture and Agri-Food Canada research farms at Saskatoon and Scott, Saskatchewan, Canada and Beaverlodge, Alberta, Canada. Soil fertility was optimized for canola production in late fall (Beaverlodge) or early spring (Saskatoon, Scott) prior to sowing the plots; trial sites also were treated with ethalfluralin (5% granular, 20.0 kg ha 1 fall applied, Saskatoon) or trifluralin (5% granular, 22.3 kg ha 1 fall applied, Scott; 480 g L 1 emulsifiable concentrate, 2.5 L ha 1 spring applied, Beaverlodge) prior to sowing. One trial was sown at each location in 2002; in 2005, one trial was sown at Scott and two trials were sown about 2 wk apart (mid- and late May) at Saskatoon and Beaverlodge. Details of the experimental design, seeding, pest control and harvest were as described for 2001 except that plot dimensions differed among locations due to the configuration of seeding equipment and to pre-harvest trimming of plots so that all plots at a location were the same length. Plots at Saskatoon and Scott consisted of four rows spaced 30 cm apart; plots at Beaverlodge consisted of six rows spaced 20 cm apart. Planted row length was 6 m (Saskatoon), 7 m (Scott) or 8 m (Beaverlodge). Plots were desiccated with diquat (240 g L 1 solution, 2.0 L ha 1 ) + Agral 90 (1 ml L 1 spray solution) prior to harvest except at Saskatoon in Harvested plot area for each location was 7.32 m 2 (Saskatoon), 6.15 m 2 (Scott) and 7.20 m 2 (Beaverlodge). Data Collection All data were recorded on a plot basis. Days to flower was determined as the number of days from date of seeding to approximately 20% of the plants in a plot having open flowers. Days to maturity was determined as the number of days from date of seeding to physiological or swathing maturity, i.e., when about 95% of the pods had changed colour and the seeds were firm, representing a moisture content of about 25%. Crop height (cm) was the mean of two or three measurements made near the centre of each plot after flowering and prior to swathing maturity. Lodging severity was determined after flowering and prior to swathing maturity using a 1 (plants erect) to 5 (plants flat on ground) visual rating scale. To determine seed yield, plots were harvested separately at full plant maturity using a Hege 140 plot combine; seed was then dried to approximately 7% moisture content, cleaned, weighed (g) and g plot 1 values converted to kg ha 1 based on plot dimensions for each location. Thousandseed weight was measured on a randomly selected sample of 500 counted seeds dried to < 5% moisture content and expressed as grams per 1000 seeds. Oil content, reported as percent whole seed on a dry weight (zero moisture) basis, was determined on 20.0 g of seed using a Newport nuclear magnetic resonance spectrometer [American Oil Chemist Society (AOCS) 2000, Recommended Practice Ak 3-94]. Protein content, reported as percent whole seed on a dry

4 1050 CANADIAN JOURNAL OF PLANT SCIENCE weight basis, was determined on 0.5 g of seed by the Dumas Combustion method (AOCS 1999, Official Method Ba 4e- 93); total nitrogen content of the seed was measured and the value was multiplied by a correction factor of 6.25 to give a value for protein. Oil and protein contents of seed harvested in 2005 were determined on 5.0 g of whole seed using nearinfrared reflectance (NIR) spectroscopy (FOSS NIRSystems model 6500 spectrometer, spinning cup autosampler, WIN ISI II calibration software); NIR calibration sets were developed by Dr. J. P. Raney, Agriculture and Agri-Food Canada, Saskatoon. Fatty acid composition of the seed oil was determined by the method of Thies (1971), except that gas chromatography of the fatty acylmethyl esters was performed with a 0.32 mm (internal diameter) 0.5 µm (phase thickness) 15 m (length) HP-Innowax fused silica capillary column. The methyl esters were prepared by base-catalyzed transesterification using sodium methoxide in methanol and hexane on oil extracted from 2.0 g of crushed seed. Individual fatty acids are reported as mass percent of total fatty acids, and averages are the arithmetic mean of all available data. Data Analysis Yield data were analyzed using the analysis of variance procedure SAS ANOVA (SAS Institute, Inc ) for a randomized complete block design. The coefficient of variation (CV, %) and least significant difference (LSD, P < 0.05) also were determined for each trial. For the 2002 and 2005 trials, Bartlett s test was used to group environments according to homogeneity of residual variances. Species means were calculated for each trait; ranges of estimates also were calculated for C. sativa. RESULTS All trials were harvested except the trial at Scott in 2005, which was abandoned because of severe hail and flood damage. Also, two replicates in the early-seeded trial at Beaverlodge in 2005 (Test 1) were abandoned because of poor establishment caused by a severe infestation of weeds. The 2001 growing season at Saskatoon was generally warmer and drier than normal (Table 2). Good germination and uniform stand establishment were noted for all C. sativa and Brassica plots in early June, but persistent drought conditions throughout the growing season accelerated maturity and reduced yields. Similarly, a severe moisture deficit prior to spring sowing and warmer than normal temperatures in June and July were characteristic of the 2002 growing season across much of western Canada. No rain was recorded at Saskatoon and Scott in May when the trials were planted, nor did any significant amount of rain fall at these locations until late June (Table 2). Furthermore, Scott remained relatively dry until August, resulting in below normal total precipitation, but Saskatoon was wetter than normal in July and August, resulting in normal total precipitation for the May to August growing season. Total precipitation at Beaverlodge also was well below normal for this period, but the spring moisture deficit was not as severe as at the other two locations. Temperatures in June and July were warmer than normal at all three locations (Table 2). The widespread drought caused poor emergence and uneven stand establishment in many commercial crops in western Canada in 2002, although research plots fared somewhat better due to the intense management of these sites. On average, the C. sativa plots at Saskatoon and Scott germinated well and were more uniform in establishment than the Brassica plots; among the Brassica species in the trials, B. juncea was the most vigorous and uniform, followed by B. napus and B. rapa. In contrast, the 2005 growing season was generally cooler and wetter than normal at Saskatoon and cooler with near normal precipitation at Beaverlodge. Nearly three times the normal amount of rain fell at Saskatoon in June following a relatively dry May, and nearly twice the normal amount fell in August (Table 2). Germination of all crop species was good in both early- and late-seeded trials at Saskatoon and Beaverlodge, but stand establishment was more uniform in the later-seeded trials probably because of more competition from weeds like stinkweed (Thlaspi arvense L.) in the earlier-seeded trials. No obvious differences in stand establishment were noted among crop species. Yield data from each trial were analyzed separately and the error mean squares examined for consistency across trials. Bartlett s test for homogeneity of variances (P < 0.05) indicated that residual variances for the seven trials with all 19 C. sativa accessions conducted in 2002 and 2005 were not homogeneous. Analysis of variance on yield for individual trials revealed significant effects of replicate and entry for all trials except Saskatoon (2001), Scott (2002) and Beaverlodge Test 2 (2005), where replicate was not significant (Table 3). Coefficients of variation for yield ranged from 5.4% (2005 Saskatoon Test 2) to 21.4% (2002 Saskatoon) (Table 3). Mean yields for individual trials are reported in Table 4. Residual variances for some combinations of trials were similar, which allowed some combinations of trials within years and locations to be pooled to examine the effects of location and year, respectively, on yield. For these analyses year, location and replicate were treated as random effects and entry was treated as a fixed effect. The pooled data from the Scott and Beaverlodge trials in 2002 (χ 2 = 1.55, CV = 9.3%) and the Saskatoon (Test 1) and Beaverlodge (Test 2) trials in 2005 (χ 2 = 2.28, CV = 8.2%) revealed significant effects of location and replicate within location; entry was significant only for the 2002 pooled trials and the location entry interaction was significant only for the 2005 pooled trials (Table 5). The pooled data from the Beaverlodge trials in 2002 and 2005 (Test 2) (χ 2 = 3.42, CV = 7.2%) revealed significant effects of year, replicate within year and year entry (Table 5). Mean values for agronomic and seed quality traits are reported in Tables 6 and 7. Means from the preliminary trial at Saskatoon in 2001 are not reported since only four C. sativa accessions were tested that year. For C. sativa, mean days to flower was similar (40 43 d) in all trials except the earlyseeded trials in 2005, where mean days to flower was greater (47 49 d). Camelina sativa matured 3 to 4 d later than B. rapa in 2002 and up to 6 d later in 2005; maturity was latest at Beaverlodge in both years. Brassica napus was the last to flower and mature in all trials. At Beaverlodge in 2002, B. napus was not mature by mid-september when the

5 GUGEL AND FALK CAMELINA SATIVA IN WESTERN CANADA 1051 Table 2. Environmental conditions at Saskatoon and Scott, Saskatchewan and Beaverlodge, Alberta during the 2001, 2002 and 2005 growing seasons compared to long term normals Mean air temperature Total precipitation LTN z LTN Location (soil type) Month ( C) (mm) Saskatoon (clay) May N June W July August y Scott (clay loam) May N June W July August Beaverlodge (silty clay) May N June W July August z Long Term Normal. y Total precipitation for growing season (May 01 Aug. 31). Table 3. Analysis of variance for seed yield of 19 accessions of Camelina sativa and one accession each of Brassica rapa, B. juncea and B. napus grown in field trials at Saskatoon and Scott, Saskatchewan and Beaverlodge, Alberta in 2001, 2002 and 2005 z trial was desiccated (day 112); similarly, in 2005 both B. napus and B. juncea were not mature by mid-september when the trial sown at Beaverlodge in late May (Test 2) was desiccated (day 111). The B. napus plots at Beaverlodge in 2002 and 2005 (Test 2 only) also were severely damaged by frost, while the B. juncea plots were affected to a lesser degree; no frost damage was noted for B. rapa or C. sativa in any trial. Crop height did not vary by more than 8 cm among C. sativa accessions in the 2002 trials, but in 2005 crop height was more variable especially in the early-seeded trials. All C. sativa accessions were shorter than the Brassica checks in all trials, and all accessions were taller in 2005 than in Little or no lodging was observed in any trial, so the data are not reported. In all trials, 1000-seed weight was lower for C. sativa ( g) than for any of the Brassica checks ( g across all species). The lowest yields of all crop species were observed at Scott in 2002; yields were generally highest in the laterseeded trials at Saskatoon and Beaverlodge in 2005 (Tables 4, 6 and 7). Most C. sativa accessions had higher yields than Mean squares y Source df 01SK x 02SK 02SC 02BL 05SK1 05SK2 05BL1 w 05BL2 Entry *** *** *** *** *** *** ** *** Replicate NS *** NS *** *** *** *** NS Residual error CV (%) z 2001 field trial: 01SK = Saskatoon; 2002 field trials: 02SK = Saskatoon, 02SC = Scott, 02BL = Beaverlodge; 2005 field trials: 05SK1 = Saskatoon Test 1, 05SK2 = Saskatoon Test 2, 05BL1 = Beaverlodge Test 1, 05BL2 = Beaverlodge Test 2. y Significant at P < 0.01 (**), P < (***) or not significant (NS). x Preliminary trial with four accessions of C. sativa and one accession each of B. rapa, B. juncea and B. napus: Entry (6 df), Residual error (18 df). w Two replicates only: Replicate (1 df), Residual error (21 df). any of the Brassica checks in the 2002 trials (Tables 4 and 6). In 2005, however, B. juncea had the highest yield in all but the early-seeded trial at Beaverlodge (Test 1), where B. napus had the highest yield. Most C. sativa accessions had higher yields than B. rapa at Saskatoon in 2005, but the reverse was true at Beaverlodge (Tables 4 and 7). Yields of B. napus at Beaverlodge in 2005 are understated due to losses from shattering of about 10 and 25% in Tests 1 and 2, respectively; very minor losses (< 1%) from shattering also were noted for the earliest maturing C. sativa accessions in Test 2. Average seed oil contents of C. sativa and the Brassica checks ranged from 38 to 43% and from 38 to 53%, respectively, across all trials (Tables 6 and 7). Oil contents of all crop species tested were highest at Beaverlodge in 2002 and 2005 and higher overall in 2005 than in Within a trial, B. napus and B. rapa generally had similar oil contents, while B. juncea always had the lowest oil content of the Brassica checks. This was not surprising since condiment B. juncea has been extensively selected for low oil content. On

6 1052 CANADIAN JOURNAL OF PLANT SCIENCE Table 4. Mean yields of Camelina sativa, Brassica rapa, B. juncea and B. napus accessions grown in field trials at Saskatoon and Scott, Saskatchewan and Beaverlodge, Alberta in 2001, 2002 and 2005 z Yield 01SK 02SK 02SC 02BL 05SK1 05SK2 05BL1 05BL2 Accession (kg ha 1 ) CS CS CS CS CS CS CS CS CS CS CS CS CS CS CS CS CS CS CS BR01 (AC Parkland) BJ01 (AC Vulcan) BN01 (AC Excel) LSD z 2001 field trial: 01SK = Saskatoon; 2002 field trials: 02SK = Saskatoon, 02SC = Scott, 02BL = Beaverlodge; 2005 field trials: 05SK1 = Saskatoon Test 1, 05SK2 = Saskatoon Test 2, 05BL1 = Beaverlodge Test 1, 05BL2 = Beaverlodge Test 2. Table 5. Analysis of variance of pooled data for seed yield of 19 accessions of Camelina sativa and one accession each of Brassica rapa, B. juncea and B. napus grown in field trials at Saskatoon and Scott, Saskatchewan and Beaverlodge, Alberta in 2002 and 2005 z Mean squares y Source df 02SC + 02BL 05SK1 + 05BL2 02BL + 05BL2 Entry *** NS NS Year *** Replicate(Year) ** Year Entry *** Location *** *** Replicate(Location) ** *** Location Entry NS *** Residual error χ NS 2.28 NS 3.42 NS CV (%) z 2002 field trials: 02SC = Scott, 02BL = Beaverlodge; 2005 field trials: 05SK1 = Saskatoon Test 1, 05BL2 = Beaverlodge Test 2. y Significant at P < 0.01 (**), P < (***) or not significant (NS). For the location effect, error terms for Entry and Location were Location Entry and Replicate(Location), respectively; for the year effect, error terms for Entry and Year were Year Entry and Replicate(Year), respectively. average, the oil content of C. sativa in a trial was always less than that of B. juncea, although the ranges of values observed for the pooled C. sativa accessions in some trials (2002: Saskatoon and Scott; 2005: Saskatoon Test 1) indicated that some individual accessions had potentially higher oil contents than B. juncea in some environments. Protein content was negatively correlated with oil content (r = 0.91 to 0.98); for each crop species and across all trials, the highest oil contents were associated with the lowest protein contents. Average protein contents of C. sativa and the Brassica checks ranged from 27 to 32% and from 21 to 33%, respectively, across all trials. Brassica juncea had the highest protein content of the Brassica checks and B. rapa the lowest; the protein content of C. sativa was generally slightly less than that of B. juncea in 2002, but the reverse was true in Wider ranges of values for fatty acid composition of C. sativa seed oil were observed among accessions within a trial than among locations in 2002 and 2005 (Tables 6 and 7). Fatty acid composition of C. sativa across all trials was

7 GUGEL AND FALK CAMELINA SATIVA IN WESTERN CANADA 1053 Table 6. Mean values for agronomic and seed quality traits of 19 accessions of Camelina sativa and one accession each of Brassica rapa, B. juncea and B. napus grown in field trials at Saskatoon and Scott, Saskatchewan and Beaverlodge, Alberta in 2002 Saskatoon Scott Beaverlodge Camelina sativa Brassica spp. means Camelina sativa Brassica spp. means Camelina sativa Brassica spp. means Trait Mean SD z Range rapa juncea napus Mean SD Range rapa juncea napus Mean SD Range rapa juncea napus Days to flower Days to maturity y Crop height (cm) seed weight (g) Seed yield (kg ha 1 ) % Oil (dry wt. basis) % Protein (dry wt. basis) Fatty acids (% of total): C18:1 - oleic C18:2 - linoleic C18:3 - linolenic C20:1 - eicosenoic C20:2 - eicosadienoic Tr x 1.2 Tr Tr 1.2 Tr Tr 1.1 Tr C20:3 - eicosatrienoic Tr Tr Tr Tr Tr Tr Tr Tr Tr C22:1 - erucic Tr 24.9 Tr Tr 25.1 Tr Tr 23.3 Tr Total saturated w Other fatty acids v z SD = standard deviation of the mean of 19 C. sativa accessions. y Plots not mature on day 112 when the trial was desiccated to facilitate harvest. x Tr = Trace (included in Other fatty acids ). w Includes C12:0 (lauric), C14:0 (myristic), C16:0 (palmitic), C18:0 (stearic), C20:0 (arachidic), C22:0 (behenic) and C24:0 (lignoceric). v Includes C16:1 (palmitoleic), C22:2 (docosadienoic), C22:3 (docosatrienoic), C24:1 (nervonic) and trace amounts of other unidentified fatty acids.

8 1054 CANADIAN JOURNAL OF PLANT SCIENCE Table 7. Mean values for agronomic and seed quality traits of 19 accessions of Camelina sativa and one accession each of Brassica rapa, B. juncea and B. napus grown in field trials at Saskatoon, Saskatchewan and Beaverlodge, Alberta in 2005 Saskatoon Test 1 Saskatoon Test 2 Camelina sativa Brassica spp. means Camelina sativa Brassica spp. means Trait Mean SD z Range rapa juncea napus Mean SD Range rapa juncea napus Days to flower Days to maturity Crop height (cm) seed weight (g) Seed yield (kg ha 1 ) % Oil (dry wt. basis) % Protein (dry wt. basis) Fatty acids (% of total): C18:1 - oleic C18:2 - linoleic C18:3 - linolenic C20:1 - eicosenoic C20:2 - eicosadienoic Tr x 1.1 Tr Tr 1.1 Tr C20:3 - eicosatrienoic Tr Tr Tr Tr Tr Tr C22:1 - erucic Tr 25.2 Tr Tr 25.5 Tr Total saturated w Other fatty acids v Beaverlodge Test 1 Beaverlodge Test 2 Camelina sativa Brassica spp. means Camelina sativa Brassica spp. means Trait Mean SD Range rapa juncea napus Mean SD Range rapa juncea napus Days to flower Days to maturity y Crop height (cm) seed weight (g) Seed yield (kg ha 1 ) % Oil (dry wt. basis) % Protein (dry wt. basis) Fatty acids (% of total): C18:1 - oleic C18:2 - linoleic C18:3 - linolenic C20:1 - eicosenoic C20:2 - eicosadienoic Tr 1.2 Tr Tr 1.2 Tr C20:3 - eicosatrienoic Tr Tr Tr Tr Tr Tr C22:1 - erucic Tr 24.6 Tr Tr 24.3 Tr Total saturated Other fatty acids z SD = standard deviation of the mean of 19 C. sativa accessions. y Plots not mature on day 111 when the trial was desiccated to facilitate harvest. x Tr = Trace (included in Other fatty acids ). w Includes C12:0 (lauric), C14:0 (myristic), C16:0 (palmitic), C18:0 (stearic), C20:0 (arachidic), C22:0 (behenic) and C24:0 (lignoceric). v Includes C16:1 (palmitoleic), C22:2 (docosadienoic), C22:3 (docosatrienoic), C24:1 (nervonic) and trace amounts of other unidentified fatty acids.

9 GUGEL AND FALK CAMELINA SATIVA IN WESTERN CANADA 1055 primarily oleic ( %), linoleic ( %), linolenic ( %) and eicosenoic ( %), with lesser amounts of erucic ( %) and other unsaturated fatty acids; saturated fatty acids constituted 8.7 to 9.6% of the total fatty acids and included primarily palmitic ( %) and stearic ( %) acid. Total saturated fatty acid content for C. sativa was lowest at Beaverlodge in both years ( %). DISCUSSION Seeding equipment used to sow canola and mustard research plots needed no modification to sow C. sativa. Similarly, small plot combine harvesters easily threshed the small seed of C. sativa without significant losses once appropriate adjustments to air flow were made. In other agronomic trials, sowing the crop with different types of seeding equipment revealed no major problems (Robinson 1987), but some difficulties in emergence after direct seeding were observed in trials conducted in Italy and these were ascribed to small seed size (Angelini et al. 1997). Mechanical harvesting with standard equipment also posed no problems, although reduced air flow and regular monitoring were recommended (Bonjean and Le Goffic 1999). Zubr (1997) also noted that an ordinary combine harvester adjusted for harvesting rapeseed could be used for C. sativa. Thus, the seeding and harvesting equipment typically used on commercial farms in western Canada are probably suitable for C. sativa. Brassica rapa and B. napus were chosen as checks since both oilseed species are currently grown commercially in all three environments that were part of this study. Brassica juncea also was included since its maturity is often intermediate between B. rapa and B. napus; B. juncea is not, however, currently grown commercially in the Peace River region of northern Alberta. Plessers et al. (1962) used flax (Linum usitatissimum L.) in addition to B. rapa and B. napus checks in their study of C. sativa in northern Alberta and noted that flax matured more than 5 wk later than B. rapa and nearly 2 wk later than B. napus in this environment. The growing season in northern Alberta is relatively short and early frosts are common, which often causes severe reductions in yield and quality of later maturing crops; flax was not, therefore, included in the present study. Moderate damage to plants from flea beetle feeding was evident in the B. napus and B. rapa plots at Saskatoon and, to a lesser degree, in the B. juncea plots even after multiple applications of insecticide prior to flowering. Less damage, and fewer flea beetles, were observed in 2005 than in 2001 and In contrast, no feeding damage was noted in the C. sativa plots even though flea beetles were observed on the plants. High resistance to flea beetle feeding has been reported for C. sativa (Pachagounder et al. 1998; Soroka et al. 2003), although the resistance may have resulted from the absence of cues that initiate feeding rather than a feeding deterrent (Pachagounder et al. 1998). Blackleg disease [Leptosphaeria maculans (Desmaz.) Ces. & de Not.] was present at low levels (< 1% incidence) in the canola plots each year, but none was observed in the C. sativa plots. A few plots of all crop species with < 1% incidence of aster yellows disease also were noted in 2001 and 2002; howev- er, in 2005 most C. sativa plots had up to 3% incidence of aster yellows, but only one B. juncea plot in Test 2 was affected. More research is needed to determine whether the leafhopper vector (Macrosteles quadrilineatus Forbes) of the aster yellows phytoplasma feeds preferentially on C. sativa. No attempt was made to determine actual losses in yield due to insects and diseases. Observations on insect and disease resistance were not recorded at Scott and Beaverlodge. Variability in yield was relatively high at Saskatoon in 2002, although no differences among replicates in stand establishment or plant density were obvious from visual inspection of the trial. The variability may have been due in part to small differences in soil fertility, soil moisture or both across the plot area, differences that were likely exacerbated by the prevailing drought conditions and not accounted for in the experimental design. The data do, nevertheless, support the general trends observed in the other trials and are included here for the record. Seed yields of the C. sativa and Brassica accessions were influenced by location (Tables 4, 5, 6 and 7). Similar effects of location on yield of these oilseed species in Canada were reported by Plessers et al. (1962), who conducted field trials from 1958 to 1960 at Fort Vermilion, Alberta and Ottawa, Ontario. More recent trials conducted in Austria (Vollmann et al. 1996), Australia (Francis and Campbell 2003), Germany (Marquard and Kuhlmann 1986), the United States (Putnam et al. 1993) and jointly in Canada and Germany (Paul et al. 2000) confirm these findings. Yields of C. sativa in our trials were generally higher than those of the Brassica oilseeds tested when drought conditions prevailed (2002: Saskatoon and Scott). This finding, and the more uniform plot establishment of C. sativa in 2002, indicates that C. sativa tolerated the drought conditions much better than the Brassica oilseeds. Porcher (1863) and Zubr (1997) also reported that C. sativa is relatively resistant to drought, probably because the species is able to compensate for early water deficits (Bramm et al. 1990). However, the Brassica oilseeds were often higher yielding than the C. sativa accessions in our trials when rain was more timely and abundant. Similarly, Paul et al. (2000) grew one replicate each of 10 accessions of C. sativa and two B. napus canola cultivars at Edmonton, Alberta in 1996 and found that the canola cultivars Westar and Quantum had higher yields than the C. sativa accessions; in that year, near or above normal rainfall was recorded at Edmonton (Environment Canada, data not shown). Thus, drought was likely a major factor in reducing the yields of the Brassica oilseeds relative to those of the C. sativa accessions in some trials. Furthermore, the effects of late maturity also reduced the yields of B. napus and, to some extent, B. juncea in the later-seeded trials at Beaverlodge since frost and desiccation late in the growing season prevented these species from realizing their full yield potential. In general, the yield of C. sativa has been competitive with the yields of Brassica oilseeds (Plessers et al. 1962; Robinson 1987; Putnam et al. 1993; Francis and Campbell 2003). Losses in yield of C. sativa caused by shattering (Plessers et al. 1962) or fruit dehiscence (Angelini et al. 1997) was not a significant problem in our trials.

10 1056 CANADIAN JOURNAL OF PLANT SCIENCE The small seed size of C. sativa may be a deterrent to immediate adoption of this oilseed in western Canada regardless of its positive attributes since small-seeded crops are generally more difficult to manage with large agricultural equipment. Small seed size may also negatively influence the efficiency of oil extraction from C. sativa (Vollmann et al. 1996). The C. sativa accessions with the largest seed in our trials (CS01, CS16) consistently produced seed ranging from 1.5 to 1.8 g per 1000 seeds; in contrast, the smallest Brassica seed (B. rapa, BR01) was in the range of 2.1 to 2.8 g per 1000 seeds (Tables 6 and 7). Seed size of C. sativa is highly heritable (Marquard and Kuhlmann 1986; Vollmann et al. 1996) and accessions with larger seed have been developed, but with a corresponding decline in oil content and seed yield (Vollmann et al. 1996). However, in a similar study on spring oilseed rape Engqvist and Becker (1993) observed no correlations between yield and other characters consistent over years and crosses, and reported positive correlations between seed weight and oil content, and seed weight and yield, for some crosses. Furthermore, in our trials the accessions with the largest seed were not necessarily those with the lowest oil content. Clearly, a selection program directed toward increasing oil content, seed size and seed yield in C. sativa will require careful screening of potential parents for crossing. Variation in oil and protein content of C. sativa has been reviewed by Zubr (2003). Wide variations in these traits were ascribed to the effects of genotype and environment and some trends were identified regarding the expression of specific quality attributes at certain locations. For example, seed from Sweden tended to be high in oil content, and seed from some locations in Denmark, Germany and Ireland tended to be high in protein. Vollmann et al. (1996) also reported significant genotype, year and location effects on oil content. Results from our trials are in general agreement with previously published data on oil and protein content of C. sativa. In our trials, oil content ranged from approximately 36 to 43% at Saskatoon, 36 to 40% at Scott and 40 to 45% at Beaverlodge; similarly, protein content ranged from approximately 28 to 33% at Saskatoon, 30 to 33% at Scott and 25 to 30% at Beaverlodge (Tables 6 and 7). Plessers et al. (1962) also reported higher oil contents for C. sativa grown in northern Alberta; the Brassica oilseeds also tended to have higher oil contents when grown at northern latitudes (Tables 6 and 7; Plessers et al. 1962). To our knowledge, the C. sativa accessions evaluated in the present study are largely unselected for oil content. Since oil content is highly heritable in this species (Vollmann et al. 1996), and germplasm with higher oil content has been selected from other segregating populations (Vollmann et al. 1996), it should be possible to develop adapted populations that consistently produce seed with even higher oil contents than observed at Beaverlodge in this study. Variation among C. sativa accessions in fatty acid composition of the seed oil is reported to be narrow, but statistically significant (Budin et al. 1995; Crowley and Fröhlich 1998; Zubr and Matthäus 2002); Bonjean and Le Goffic (1999), however, report much greater variation among some C. sativa breeding lines. The variability is largely due to the effects of environment and genotype (Crowley and Fröhlich 1998; Zubr and Matthäus 2002; Francis and Campbell 2003). Zubr and Matthäus (2002) determined that the effects of environment were greater than those of genotype, and Francis and Campbell (2003) determined that the effect of genotype was greater. In our trials, more variation (wider ranges of values) was observed among accessions within a trial than among locations (Tables 6 and 7). Variation in individual fatty acids was generally small, not exceeding 4.5% in absolute terms among accessions (Table 7: linoleic acid, 2005 Saskatoon Test 1) and 2.0% among locations (Table 7: linolenic acid). However, the values reported are averages of many plants per accession and more variation may be present among individual plants within an accession than is evident from the plot data. In general, our data for individual fatty acids are in agreement with Plessers et al. (1962), Crowley and Fröhlich (1998) and Zubr and Matthäus (2002). Overall, the potential to cultivate C. sativa in western Canada is promising. In our study, the crop grew successfully at all trial sites tested, tolerated drought conditions and was unaffected by moderate to severe infestations of flea beetles. Planting and harvesting the crop were easily accomplished with equipment currently used for other small-seeded crops like canola and mustard. All the C. sativa accessions tested were uniform in height, matured relatively early and some had yields competitive with those of the Brassica oilseeds. Other positive attributes of C. sativa include a favourable response to low-input farming practices (Putnam et al. 1993; Schuster and Friedt 1995; Zubr 1997), disease resistance (Salisbury 1987; Conn et al. 1988) and the lack of seed dormancy (Robinson 1987), which would make the control of volunteer plants in other crops following C. sativa relatively simple. Furthermore, the prospects of developing improved germplasm for particular western Canadian environments are good. However, additional trials across the region, especially those investigating stand establishment, fertility requirements and broadleaf weed control options, are needed to develop the technology to optimize production on a regional basis. The best overall performance of C. sativa in our trials was observed in northern Alberta, where the growing season is relatively short; thus, our data support the suggestion that this oilseed merits serious consideration as a potential crop for this region (Plessers et al. 1962). Even so, high-yielding cultivars with larger seed and higher oil content are needed to make this crop more attractive, all the while maintaining earliness; an economically sustainable market for the crop must also be found (Putnam et al. 1993). In addition to the traditional use of camelina oil for direct human consumption (Zubr 1997), blending camelina oil with canola oil to increase the omega- 3 content of the canola oil could also be a good selling point for C. sativa (Francis and Campbell 2003). The fatty acid profile can also be manipulated to produce an oil more suited to industrial applications (Vollmann et al. 1997; Büchsenschütz-Nothdurft et al. 1998). Ideally, a western Canadian market for the seed meal also needs to be developed; potential uses include feed for poultry, swine and ruminants (Putnam et al. 1993; Zubr 1997; Schuster and