Alfalfs Seed Production

New Zealand Journal of Agricultural Research, 1995, Vol. 38: 289-295 0028-8233/95/3803-0289 $2.50/0 © The Royal Society of New Zealand 1995

289

Effect of row spacing and sowing rate on seed production of lucerne (Medicago sativa L.) cv. Grasslands Oranga M. ASKARIAN1 J. G. HAMPTON2 M. J. HILL Seed Technology Centre Department of Plant Science Massey University Palmerston North, New Zealand Present address: Ministry of Jihad-e-Sazandagi, Tehran, Iran. Abstract The effects of row spacing (15, 30,45, and 60 cm) and sowing rate (1, 3, 6, and 12 kg/ha) on lucerne seed yield and its components were investigated at Massey University, Palmerston North, New Zealand, over two seasons. In the first year, seed yield from the 15 cm row spacing was significantly lower than that from the 30, 45, and 60 cm row spacings, whereas sowing rate had no effect on seed yield. In the second year crop, row spacings did not significantly affect seed yield, but the seed yield from the 1 kg/ha sowing rate was significantly greater than that for other sowing rates because harvestable racemes/m2 and thousand seed weight were significantly increased. Seed yield over 2 years of the experiment was highest at the 1 kg/ha sowing rate and for the 30 and 45 cm row spacings. However, there were no significant interactions between row spacing and sowing rate for seed yield. The average seed yield for all treatments was 127 and 187 kg/ha for the first and second year respectively. Neither row spacing nor sowing rate had any effect on the quality of harvested seed. Keywords alfalfa; lucerne; row spacing; seed production; seed quality; sowing rate

A940I1 Received 25 February 1994; accepted 18 May 1995 Corresponding author

INTRODUCTION Seed yield of lucerne varies from year to year because the growth and development of the seed yield components is strongly affected by the environment (Rincker et al. 1988). Plant density is known to be an important factor in seed production, because competition between and within plants affects a plant's ability to produce vegetative and reproductive material. Recommendations for optimum row spacing and sowing rate for lucerne seed production vary in the literature (Askarian 1993), and several studies have reported conflicting results (Pedersen & Nye 1962; Kowithayakorn & Hill 1982; Lovato & Montanari 1987). In New Zealand, lucerne seed has commonly been produced in rows 9 or 18 cm apart at sowing rates of 6-12 kg/ha (Wynn-Williams & Palmer 1974). However, Dunbier et al. (1983) recommended a sowing rate of 1 kg/ha and row spacing of 75 cm, although they produced no evidence to support these recommendations. Experiments conducted overseas, especially in the United States, have shown that high seed yields were obtained from lucerne sown at rates ranging from 0.5 to 2 kg/ha and in rows from 60 to 150 cm apart (Abu-Shakra et al. 1969; Goplen 1972). This study reports how factors such as sowing rate and row spacing affected the seed production of lucerne cv. Grasslands Oranga at Palmerston North over two seasons at the same site. MATERIALS AND METHODS The 1400 m2 trial site was located at Massey University, Palmerston North, New Zealand (Latitude 40°S, Longitude 175°E). The soil was a Manawatu fine sandy loam (Cowie 1974). The site had previously been in ryegrass/white clover pasture. The land was ploughed in March 1990 and then fallowed. Glyphosate (Roundup®) at 0.75 kg a.i./ha was applied in February 1991 to control weeds. The site was then rotary-hoed once before sowing in March 1991.

New Zealand Journal of Agricultural Research, 1995, Vol. 38

290

The experiment was set up using a split plot design with four main plots each replicated 4 times. The main plots were four row spacings and the subplots were four sowing rates. The row spacings were 15, 30, 45, and 60 cm, and the sowing rates were 1, 3, 6, and 12 kg/ha (Askarian & Hampton 1993). Certified breeders' seed of cv. Grasslands Oranga was sown using a cone seeder at a depth of 1.5 cm on 15 March 1991. The area of each main plot was 50 m2 and each subplot was 5 x 2 m2 (10 m2), with a 0.5 m gap between each subplot and a 3 m gap between replicates. Details of agronomic management are provided in Table 1. Pollination in both years was provided by honey bees. Two honey bee colonies were placed adjacent to the trial after flowering started on 10 January 1992, and 24 December 1992. During the period of the experiment bumble bees (Bombus spp.) were also present but the population was low. Flowering pattern was recorded from a randomly allocated 50 x 50 cm permanent quadrat in each plot in both years. Only flowers with at least one open floret and no withered florets were counted. Flowers were counted every 7 days during the flowering period (Askarian 1993). Maximum and minimum temperature and rainfall for 1991-93 and 60-year average data were recorded at a station (AgResearch Grasslands) 2 km from the trial area.

Lucerne seed yield and its components were recorded at 59 and 62 days after peak flowering in the first and second year respectively by removing all plant material from two randomly allocated 0.5 m2 quadrats per plot, when the majority of pods had turned brown (Askarian 1993). Sampling was done by using a motorised hand-piece to cut plants at ground-level and all plant material was collected and bagged. Samples were then left to air-dry for 4 weeks, after which the racemes from each sample were separated by hand. Seeds from each sample were threshed by hand-rubbing (Askarian et al. 1993) and cleaned using 1.0-1.7 mm sieves and then a Burrows portable blower set at 41.4 km/h air speed for 3 min. Pure seed weight was recorded, and seed moisture content determined (ISTA 1993). Yield per unit area was obtained from the weight of seeds from each subplot in each treatment. The number of pods per raceme was determined from 50 brown-black racemes taken randomly from each subplot just after harvesting. Seeds per pod were determined by hand-threshing of racemes. Thousand seed weight (TSW) for each treatment was also determined (ISTA 1993). Seeds from each treatment were germinationtested in early May 1992, and late April 1993 using internationally agreed methodology (ISTA 1993), with four replicates of 50 seeds.

Table 1 Agronomic management and experimental detail, 1991/92 and 1992/93. 1991/92 Site Seed source Sowing date Sowing rate Row spacing Soil type Site soil analysis

1992/93

Massey University As per 91/92 AgResearch Grasslands, N.Z. 15 Mar 1991 1,3, 6, 12 kg/ha 15,30,45,60 cm Manawatu fine sandy loam pH, 6.2; Olsen P, 15; SO4, 10; Exch K, 0.27; Exch Ca, 5.56; Exch Mg,1.5; Exch Na, 0.07; CEC.15 Grazing None 30 Jul Slug control Mesurol 8 kg/ha 1 Apr, 7 May None Insect control Fluvalinate 0.1 kg a.i./ha 30 Apr, 10 Dec as per 91/92, applied 20 Nov, 15 Dec Herbicide 2,4-DB 2.4 kg a.i./ha 21 May, Propyzamide 0.75 kg a.i./ha 21 Jul Hexazinone 1.0 kg a.i./ha 1 Oct Disease control Benlate 0.25 kg a.i./ha 10 Apr, 10 Dec as per 91/92, applied 15 Dec Hand weeding Oct and late Nov 1 Nov Defoliation 8 Nov None N fertiliser 50 kg N/ha 20 Nov 25 kg N/ha 15 Nov First flowering 31 Dec 8 Dec End of flowering 20 Mar 17 Mar Pollinators introduced 10 Jan 24 Dec Seed harvest 20 Mar 17 Mar

Askarian et al.—Seed production of lucerne Statistical analysis was done using the SAS system (SAS 1991). Least significant differences at the 5% probability level (P < 0.05) were used to differentiate treatment means where analyses of variance (ANOVA) or general linear model (GLM) were significant at the 0.05 or 0.01 level of confidence.

291 are presented separately for row spacing and sowing rate.

Meteorological data Climate data for the cropping seasons 1991/92, and 1992/93 are presented in Table 2. The total amount of rain for the first season was higher than for the second season (1067.5 versus 986.6 mm, respectively). During the second cropping season, December was particularly wet (106% more than December 1991) but January and February (1993) were dryer than those months in 1992. February 1992 was also wet, having 132% more rain than the 60-year average (Table 2). Mean daily maximum and minimum spring and summer temperatures were lower by 1-2°C for the second than the first season, with an exception for November 1992. In particular, cooler conditions occurred during December, January, and February (flowering period and seed development) in the second season (1992/93), than in the first year. During these months monthly mean minimum and maximum temperatures ranged from 10.8 to 20.8°C, compared to 11.3-21.9°C in 1991/92.

1991/92 In the first season, lucerne plants started flowering on 31 December, peak flowering was recorded on 26 January, and flowering had virtually ceased at harvest on 20 March 1992, giving a flowering period of 80 days. Peak flowering date did not differ with row spacing, but the 30 cm row spacing produced significantly fewer flowers/m2 at peak flowering than the other row spacings (580 flowers/ m2 cf. 740 flowers/m2), whereas the greatest number of flowers was always produced at the 60 cm row spacing (data not presented). At final harvest, plants grown at the narrowest and widest row spacings had produced significantly fewer harvestable racemes/m2 than the two middle row spacings (Table 3). Pods per raceme did not differ among the 15, 30, and 45 cm row spacings, but there were fewer at the 60 cm than at the 30 cm row spacing (Table 3). Plants grown at the 15 cm row spacing produced significantly more seeds per pod than those at the two highest row spacings while the narrowest row spacing produced significantly smaller seeds than the other row spacings (Table 3). Plants grown at the 15 cm row spacing had a significantly lower seed yield than those from the 30 and 45 cm row spacings (Table 3), but there were no significant differences among the other row spacings.

Row spacing Because with one exception (seeds per pod in 1992/ 93) there were no significant interactions between row spacing and sowing rate (Askarian 1993), data

1992/93 Flowering in the second season started about 3 weeks earlier than in the first season, the first flower of the season being recorded on 8 December.

RESULTS

Table 2 Sixty-year average for temperature (minimum and maximum), and rainfall at Palmerston North, and deviations from these averages during 1991/92a and 1992/93. Aug

Sep

Oct

Nov Dec Jan Feb Mar Temperature (°C) Min Max Min Max Min Max Min Max Min Max Min Max Min Max Min Max 60-year ave. 5.0 13.1 6.6 14.7 8.3 16.6 9.8 18.5 11.6 20.6 12.8 21.9 12.8 22.3 11.7 20.9 1991/92 1.1 -0.2 -0.6 -1.5 -2.3 -0.3 -1.3 0.1 0.0 -0.4 -1.4 -2.2 -2.3 1.7 1.2 0.7 1992/93 -1.4 -1.1 -0.3 -1.4 -0.7 -0.6 1.2 0.0 -0.8 -1.3 -1.3 -1.0 -1.1 -1.5 -1.7 -1.7 Rainfall (mm)i 60-year ave. 89.0 75.0 88.0 78.0 94.0 79.0 67.0 69.0 1991/92 9.7 -9.9 -7.1 3.0 -12.8 -1.8 88.2 19.9 1992/93 21.1 13.0 -3.0 -17.0 73.0 -25.0 -23.0 11.0 a

Data (obtained from AgResearch Grasslands) were recorded at a station 2 km from the trial area.

New Zealand Journal of Agricultural Research, 1995, Vol. 38

292

Peak flowering occurred on 19 January, and plants were harvested on 17 March 1993. The flowering period was therefore 99 days. Differences in climatic conditions (Table 2), particularly higher temperatures in November 1992 and lower temperatures in December 1992 and January and February 1993, probably contributed to this variation. The date of peak flowering did not differ among treatments, but flower production was greater than in the first year (900 flower/m2 at peak flowering cf. 700 flowers/m2). There were some significant differences among treatments; for example, on 3 and 10 February, plants grown at the 60 cm row spacing had 100 more flowers/m2 (P < 0.05) than those in the 15 cm row spacing. Row spacing did not significantly affect seed yield or its components (Table 3), with the exception of pods per raceme which tended to decrease as row spacing increased. When seed yield data for the two seasons were meaned the 15 cm row spacing produced significantly less seed (91 kg/ha, P < 0.05) than the three other row spacings (166,177, and 149 kg/ha). Sowing rate 1991/1992 Flowering pattern was not affected by sowing rate as the date of peak flowering (28 January) was similar at all sowing rates and none of the sowing

rates increased flower numbers significantly at peak flowering (data not presented). At final harvest the 12 kg/ha sowing rate had a significantly lower number of harvestable racemes/m2 compared to the lowest seeding rate (1 kg/ha), but there were no significant differences among the first three sowing rates (Table 4). Sowing rate had no significant effect on pods per raceme, seeds per pod, or seed yield, although smaller seeds (P < 0.05) were produced from the 12 kg/ha sowing rate (Table 4). 1992/93 There were more flowers at peak flowering in the second year (950 cf. 760/m2), but flowering pattern was not affected by sowing rate. The 1 and 3 kg/ha sowing rates had 150 more flowers/m2 at peak flowering than the two highest sowing rates, but flowering pattern was not affected by sowing rate (data not presented). At final harvest the 12 kg/ha sowing rate had a significantly lower number of harvestable racemes/ m2 compared to the 1 kg/ha sowing rate. There were no significant differences among the first three sowing rates (Table 4). Pods per raceme and seeds per pod did not differ with sowing rate but there was a significant interaction (P < 0.05) for seeds per pod between the 1 and 3 kg/ha sowing rates and the 15 and 30 cm row spacings. Thousand seed weight decreased with increasing sowing rate

Table 3 Effect of row spacing on seed yield and its components, 1991/92 and 1992/93. Row spacing (cm) 1991/92 15 30 45 60 LSD P < 0.05

cv% 1992/93 15 30 45 60 LSD P < 0.05

cv%

Pods/ raceme

Seeds/ pod

764.0 980.0 995.0 840.0 136.3 21.0

5.1 5.4 5.0 4.6 0.51 14.0

3.4 3.1 3.0 3.0 0.26 11.0

1.51 1.62 1.62 1.61 0.063 5.6

102.3 142.7 144.0 120.0 29.23 22.0

1163.0 1305.0 1416.0 1381.0 NS 23.8

5.2 5.4 4.6 4.2 0.77 13.5

3.0 2.9 3.0 3.0 NS 12.0

1.74 1.75 1.73 1.79 NS 5.7

171.0 189.5 209.4 177.9 NS 20.5

TSW = thousand seed weight NS = not significant.

TSW (g)

Seed yield (kg/ha)

Harvestable racemes/m2

Askarian et al.—Seed production of lucerne

293

(Table 4), with the 1 kg/ha sowing rate producing heavier seed than the two highest sowing rates. The lowest sowing rate produced significantly more seed than the two highest sowing rates (Table 4). Mean seed yields for the two seasons were 181, 160, 152, and 135 kg/ha for the 1,3, 6, and 12 kg/ha sowing rate respectively, with the lowest sowing rate significantly (P < 0.05) outyielding the highest sowing rate. Effect on seed germination The viability, germination percentage, and hard seed percentage of lucerne seeds were not affected by row spacing or sowing rate (data not presented). All treatments produced seed which had a viability of over 98%, but germination of around 30% because of the presence of hard seed.

DISCUSSION Successful lucerne seed production is favoured in regions that are characterised by clear, sunny, warm summer days in combination with little or no rainfall during flowering. These climatic conditions promote good flowering of lucerne and provide an environment conducive to pollinating activity of bees, two factors essential for seed production (Rincker et al. 1988). However, Palmerston North is characterised by humid, windy, and often cloudy

conditions, an environment not suited to the activity of leaf cutter bees (E. Roberts pers. comm.), and for this reason honey bees were used as the pollinators. Rain which fell readily during flowering in both years (Table 2) did not encourage either lucerne flowering or bee activity (Rincker et al. 1988). Seed yields were low when compared with the 300-500 kg/ha which can be obtained in Marlborough and Canterbury (Dunbier et al. 1983), and were only 10-20% of the calculated potential seed yields of over 1000 kg/ha in both seasons (Askarian 1993). Most of this loss of potential seed yield came from poor floret site utilisation, both through poor pollination (Askarian 1993) and floret retention. Askarian et al. (1994) found that only 20% of the florets present at peak flowering were retained as pods, and from an average of 9.3 ovules per carpel at peak flowering only 3.1 seeds per pod reached maturity. Honey bees usually pollinate only 10-20% of lucerne flowers under New Zealand conditions (Forster 1974), and therefore poor pollination was not unexpected. Reasons for floret, pod, and seed abortion are not known, although assimilate shortages during ovule provisioning are likely to be involved (Askarian et al. 1994). At the same site in the 1991/92 season, the growth regulator paclobutrazol significantly increased floret retention (or conversely, decreased pod losses), but whether this was a response to altered assimilate production and/or distribution

Table 4 Effect of sowing rate on seed yield and its components, 1991 /92 and 1992/93. Row spacing (cm) 1991/92 1 3 6 12 LSD P < 0.05

cv% 1992/93 1

3 6 12 LSD P < 0.05

cv%

Racemes/m2

Pods/ raceme

Seeds/ pod

TSW (g)

Seed yield (kg/ha)

930.0 985.0 886.0 778.0 136.6 21.0

5.0 5.0 5.1 5.1 NS 14.0

3.1 3.2 3.1 3.0 NS 11.0

1.62 1.62 1.60 1.53 0.06 5.6

139.5 137.3 124.9 107.1 NS 21.9

1291.0 1155.0 1198.0 1058.0 195.77 23.0

5.4 5.2 5.1 4.9 NS 13.0

2.9 3.1 2.9 3.1 NS 12.0

1.82 1.76 1.72 1.69 0.08 5.7

223.4 182.2 178.5 163.7 44.19 20.5

TSW = thousand seed weight. NS = not significant.

294

New Zealand Journal of Agricultural Research, 1995, Vol. 38

has yet to be established (Askarian et al. 1994). This requires further investigation. In the first year, sowing rates had no significant effect on seed yield, a result which agrees with that of Sevecka (1987). In the second year however, seed yield increased as sowing rate decreased. This result generally agrees with the finding by Simko (1992) who reported that seed yield was highest at low stand densities. This higher seed yield at low sowing rates was attributed to a greater branch number (Askarian 1993) and subsequently more racemes/m2 and also a heavier seed weight. The best sowing rate for optimum seed yield was 1 kg/ha, much lower than the commonly used New Zealand rates of 5-10 kg/ha (Palmer & Donovan 1980) but similar to that recommended by Dunbier et al. (1983), and almost equal to the optimum sowing rate suggested in the United States (Rincker et al. 1988). The data in this study indicate that because of the significant reduction in yield at the 15 cm row spacing in the first year, lucerne should be planted in 30 or 45 cm rows. These spacings also facilitate inter-row cultivation (Dunbier et al. 1983). This agrees with other studies (Beran 1966; Antoniani 1972), where optimum seed yields were attained with either 45 or 30 cm between rows respectively, compared to 15 and 60 cm row spacings. Seed yield in lucerne is the product of seeds per unit area and individual seed weight (Rincker et al. 1988). Usually the greatest seed yields in herbage species are obtained from maximising seed number (Hampton & Hebblethwaite 1983). In the first year of the study the narrowest row spacing produced a lower seed yield because of fewer racemes/m2 and lower seed weight. In contrast in the second year the low sowing rate produced more racemes/m2 and heavier seed weight, because plants had had time to expand in size by developing more branches, and subsequently more racemes and seeds (Askarian 1993). These results support the findings of Dovrat (1969) and Taylor & Marble (1986). Branching may also be manipulated to increase seed yield per plant by increasing the number of primary, secondary, and tertiary lateral shoots at low sowing rates. Widely spaced plants also use less water and seeds mature more uniformly (Melton 1962). The number of seeds per pod appeared to be an unimportant yield component at different plant densities, a result also noted by Kowithayakorn & Hill (1982). The data from the lower sowing rate which produced the heaviest seed weight are consistent

with the accepted concept that individual seed weight declines with increased plant number per unit area. This was certainly the situation with the 12 kg/ha sowing rate. CONCLUSION The seed yield data recorded at this site in 1991/92 and 1992/93 provide experimental evidence for the recommendation of Dunbier et al. (1983) that lucerne for seed production in New Zealand should be sown at around 1 kg/ha, and that wide row spacings can be used. Although Dunbier et al. (1983) recommended a 75 cm row spacing, this work has shown that similar seed yields can be obtained from 30,45, and 60 cm row spacings; the decision as to which to use should therefore be dictated by the requirement for inter-row cultivation for weed control, and hence machinery design. ACKNOWLEDGMENTS Financial support for the senior author was provided by a full-time study award from the Iran Government (Ministries of Jihad-e-Sazandagi and Cultural and Higher Education), which is gratefully acknowledged. We also thank AgResearch Grasslands for the supply of lucerne seed.

REFERENCES Abu-Shakra, S; Akhar, M.; Bray, D. W. 1969: Influence of irrigation interval and plant density on alfalfa seed production. Agronomy journal 61: 562571. Antoniani, C. 1972: Trials of cultural techniques for the production of seed of lucerne (Medicago sativa L.) in the Bologna Plain. Herbage abstracts 42: 2646. Askarian, M. 1993: Seed production studies in lucerne (Medicago sativa L.) cv. Grasslands Oranga. Unpubl. PhD thesis, Massey University, New Zealand. Askarian, M.; Hampton, J. G. 1993: Effect of row spacing and sowing rate on establishment of lucerne (Medicago sativa L.) cv. Grasslands Oranga. Proceedings of the XVII International Grassland Congress: 172-173. Askarian, M.; Hampton, J. G.; Harrington, K. C. 1993: Control of weeds, and particularly white clover (Trifolium repens L.), in lucerne (Medicago sativa L.) grown for seed production. Journal of applied seed production 11: 51-55.

Askarian et al.—Seed production of lucerne Askarian, M.; Hampton, J. G.; Hill, M. J. 1994: Effect of paclobutrazol on seed yield of lucerne (Medicago sativa L.) cv. Grasslands Oranga. Journal of applied seed production 12: 9-13. Beran, V. 1966: Influence of the density of stands on the seed setting and on the yield of lucerne seed. Herbage abstracts 36: 1331. Cowie, J. D. 1974: Soils of Palmerston North City and Environs, New Zealand. New Zealand Soil Survey report 24, parts I & 2. Soil Bureau, DSIR, Wellington. Dovrat, A.; Levanon, D.; Waldman, M. A. 1969: Effect of plant spacing on carbohydrate in the root and on components of seed yield in alfalfa. Crop science 9: 33-34. Dunbier, M. W.; Wynn-Williams, R. B.; Purves, R. G. 1983: Lucerne seed production in New Zealand: achievement and potential. Proceedings of the New Zealand Grassland Association 44: 30-35. Forster, I. W. 1974: Behaviour and effectiveness of bees in pollinating legumes. Proceedings of the New Zealand Grassland Association 36: 105-110. Goplen, B. P. 1972: Management of alfalfa fields for seed production. Proceedings of the Leafcutter Bees and Alfalfa Seed Production Conference . Saskatoon. 8 p. Hampton, J. G.; Hebblethwaite, P. D. 1983: Yield components of the perennial ryegrass seed crop. Journal of applied seed production 1: 23-25. ISTA 1993: International Rules for Seed Testing. Supplement to seed science and technology 21: 1-288. Kowithayakorn, L.; Hill, M. J. 1982: A study of herbage and seed production of lucerne {Medicago sativa L.) under different plant spacing and cutting treatments in the seeding year. Seed science and technology 10: 3-12.

295 Lovato, A.; Montanari, M. 1987: Influence of row spacing and sowing rates on lucerne (Medicago sativa L.) seed production. Journal of applied seed production 5: 69. Melton, B. 1962: Effects of planting methods and seeding rates on alfalfa seed yields. Agricultural Experimental Station, New Mexico State University research report 676. Palmer, T. P.; Donovan, B. J. 1980: Seed production of new cultivars of lucerne. Pp. 87-91 in: Herbage seed production, Grassland Research and Practice Series 1, Lancashire, J. A. ed. Pedersen, M. W.; Nye, W. P. 1962: Additional factors associated with seed yield. In: Alfalfa seed production studies. Utah Agricultural Experimental Station bulletin 436. 22 p. Rincker, C. M.; Marble, V. L.; Brown, D. E.; Johansen, C. A. 1988: Seed production practice. Pp. 9851021 in: Alfalfa and alfalfa improvement, Agronomy 29, Hanson, A. A.; Barnes. D. K.; Hill, R. R. ed. Sevecka, L. 1987: The effect of sowing rate on lucerne seed yield. Herbage abstracts 57 (9): 2163. Simko, J. 1992: Effect of sowing rate, density, stand age and year on the seed yield of lucerne. Herbage abstracts 62: 3697. Taylor, A. J.; Marble, V. L. 1986: Lucerne irrigation and soil water use during bloom and seed set on a red-brown earth in S. E. Australia. Australian journal of experimental agriculture 26: 577581. Wynn-Williams, R. B.; Palmer, T. P. 1974: Seeding rates, row spacing and lucerne (Medicago sativa L.) cv. Saranac seed production. Proceedings of the Agronomy Society of New Zealand 4: 63-66.