Evaluation of new hybrid brachiaria lines in Thailand . 2 . Seed production

Forty-three new hybrid bracharia lines bred at CIAT, Colombia, were evaluated for seed production in Northeast Thailand between 2006 and 2010 in 2 experiments at 2 sites, Ubon Ratchathani and Amnart Charoen. These lines were compared with Mulato II hybrid brachiaria. From the BR02 collection, 4 lines, BR02/1718, BR02/1752, BR02/1794 and BR02/0465, were granted Plant Variety Rights in 2011. BR02/1794 produced more seed than Mulato II on most occasions, including both harvests at Ubon Ratchathani and 2 of 3 harvests at Amnart Charoen. The next best yielding lines were BR02/1718 and BR02/0465, which produced more seed than Mulato II in 1 of 2 harvests at Ubon Ratchathani and 2 of 3 harvests at Amnart Charoen. Seed-set (percentage of cleaned seed to spikelets) was generally very low in all hybrid lines (1–12%). The reasons for low seed-set in hybrid brachiaria grasses are discussed, including: being a common defect in newly formed apomictic forage hybrids; previous selection for seed yield not being rigorous enough; and insufficient selection at latitudes and sites where commercial brachiaria seed production is practiced.


Introduction
Mulato II [Brachiaria ruziziensis (now Urochloa ruziziensis) x B. decumbens (now U. decumbens) x B. brizantha (now U. brizantha)] was the second hybrid brachiaria cultivar released from the hybridization programs begun in 1988 at the Centro Internacional de Agricultura Tropical (CIAT) in Cali, Colombia (Argel et al. 2007).Even though Mulato II produced 60% higher seed yields than Mulato (Brachiaria ruziziensis x B. brizantha), which was the first hybrid brachiaria released (Hare et al. 2007a), Mulato II seed yields of 232-258 kg/ha were still very low compared with yields from other commercial brachiaria cultivars (not hybrids) elsewhere.In order to compete in price internationally with commercial brachiaria cultivars from Brazil and Australia, commercial seed yields from hybrid brachiarias must be at least 600-700 kg/ha.Commercial seed yields average 650-700 kg/ha in Brazil for cv.Marandu (B.brizantha) and cv.Basilisk (B.decumbens) (Souza 1999).In Australia, seed yields of Basilisk have reached 1,000 kg/ha (Hopkinson and Clifford 1993).Seed of these commercial brachiaria species is almost half the price of hybrid brachiaria seed.The high price of hybrid brachiaria seed is a reflection of low seed yields and represents a significant barrier to farmer uptake.
From 2006 to 2011, studies were conducted in Thailand on hybrid brachiaria collections from CIAT.The first paper of these studies reported on forage production and quality (Hare et al. 2015), while this paper focuses on seed production.

Materials and Methods
Two experiments were conducted with the aim of selecting lines that had higher seed yields than Mulato II.

Experiment 1. BR02 and MX02 collections
The first experiment was conducted at Ubon Ratchathani University, Thailand, (15° N, 104° E; 130 masl) during 2006 and 2007 alongside a forage biomass experiment.The site was on an upland sandy low humic gley soil that was acid (pH 4.6) and low in organic matter (1.1%), N (0.04%), P (3.5 ppm) and K (27.4 ppm).The mean rainfall was 1,620 mm, with a dominant dry season from November to April (Figure 1).The site is further described in the first paper on forage production (Hare et al. 2015).Thirteen hybrid brachiaria lines from the BR02 collection and 2 from the MX02 collection (Hare et al. 2015) were planted in a randomized complete block design with 3 replicates in June 2006.Seedlings were grown in a nursery and transplanted into the field plots using 50 x 50 cm spacings (48 plants per plot).Details of field crop management are presented in Table 1.Two seed harvests were conducted in 2006 and 2007.

Experiment 2. BR06 collection
This experiment was conducted at one site at the Amnart Charoen Livestock Development Centre, Amnart Charoen province, Northeast Thailand (15.5°N, 104.4°E; 168 masl) from 2008 to 2010 (3 harvests of each plot) alongside the forage trial.The site was on an upland sandy reddish brown earth with a mean rainfall of 1,640 mm, and a dominant dry season from November to April (Figure 2).Soil samples taken at planting in July 2008 showed that the soil was acid (pH 4.6), sandy (75%), and low in organic matter (0.4%), N (0.04%), and K (31 ppm), and adequate for P (25.2 ppm).The site is described further in the first paper on forage production (Hare et al. 2015).For both experiments all inflorescences in 3 m of the middle 2 rows were counted once a week.Twenty inflorescences were taken from just outside this area for reproductive analysis at peak anthesis (Table 3).All racemes were counted on each inflorescence and spikelets were counted on 3 racemes per inflorescence, selected from the top, middle and bottom of each inflorescence.At peak anthesis, nylon bags were tied over each seed head of 10 plants (5 plants/row in the above middle 2 rows) to collect the seed.The seed was allowed to fall naturally into the bags and collected once at the end of the season and cleaned through hand screens and a small seed blower to 99% pure seed.Settings were adjusted according to seed weights of each line.Following cleaning, seed yields were corrected to 10% seed moisture content.One thousand seed weights (TSW) were calculated by drying 4 lots of 100 seeds per plot and correcting to 10% seed moisture.
Data from the experiments were subjected to analysis of variance, using the IRRISTAT program from the International Rice Research Institute (IRRI).Entry means were compared using Fisher's protected LSD (P≤0.05)procedure.

Rainfall
Experiment 1. BR02 and MX02 collections.Rainfall for the experimental period is shown in Figure 1.The critical period of rainfall for seed production in Thailand is the period from July to October, when the plants establish, develop, and initiate and elongate inflorescences and seed is set and matures.The medium-term mean (13 years) rainfall at Ubon Ratchathani for this period is 917 mm, and in 2007 rainfall closely approximated the mean, but in 2006, rainfall during this critical period was 17% higher.October and November are important months for seed maturity and harvest.In 2006 and 2007, rainfall during these months exceeded the mean, by a factor of 1 in 2006 and 0.5 in 2007.Closing cuts, 5 cm above ground level, were implemented about 90 days before peak anthesis (recorded in the first year in 2008) to avoid seed head lodging prior to anthesis.
For the October-November period, rainfall in 2008 and 2009 was 40 and 80%, respectively, lower than the medium-term mean.In 2010, rainfall for the same period was 62% higher than the mean but with no rain at all during November.

Seed production
Experiment 1. BR02 and MX02 collections.Seed yields ranged from 12 to 282 kg/ha and one line produced no seed at all in the second year.Hybrid brachiaria line BR06/1794 produced significantly higher seed yields than all other lines in both years, except for BR02/1718 and BR02/0465 in the second year (Table 4).BR02/1718 and BR02/0465 also produced higher seed yields than Mulato II in the second year but not in the first year.One line, BR02/1245, failed to produce any seed in the second year.
Mulato II produced significantly lower numbers of inflorescences per m 2 than many of the other hybrid lines, which were also significantly lower than the overall mean (Table 4).Lower numbers of inflorescences per m 2 were produced in the first year compared with the second year.
Racemes per inflorescence, spikelets per raceme and TSW were lower in the second year than in the first year (Table 4).There was large variability in spikelet numbers among the lines, ranging from 24 to 48.BR02/0465 produced significantly heavier seed (10.3-10.5 g per 1,000 seeds) than all other lines (Table 4).Three lines (BR02/1485, 1747 and 1794) had significantly higher TSW than Mulato II at both harvests.
Experiment 2. BR06 collection.Seed yields ranged from 6 to 659 kg/ha with 2 lines producing no seed in some years and 1 line producing no seed at any harvest (Table 5).In the first year (2008), BR02/1794 and Marandu produced significantly more seed than the other hybrid lines, including Mulato II.The majority of the BR06 lines had lower seed yields in 2008 than Mulato II, Marandu and Toledo and the BR02 lines, except for BR02/1372, which produced low seed yields at every harvest.In the second year (2009), BR02/0465 produced a significantly higher seed yield than the other hybrid lines and cultivars.Seed yields of many of the BR06 lines improved, with BR06/1278 producing similar seed yields to Mulato II, and BR06/0423 and BR06/1000 producing, respectively, 412 and 400 kg/ha.In the third year, seed yields of nearly all cultivars and lines declined significantly, except for BR06/1000 and 2058, which produced a little over 200 kg/ha (Table 5).BR06/1254 failed to produce seed at any harvest.
The majority of the BR06 lines produced significantly higher numbers of inflorescences (300-400/m 2 ) in the first year than the cultivars and the BR02 lines (Table 5).In the second year, inflorescence numbers increased compared with numbers in the first year for most lines and cultivars, with a similar range (300-600/m 2 ) for BR02 and BR06 lines and Mulato II.In the third year, there was a substantial decrease in inflorescence numbers for all lines and cultivars, particularly for Mulato II and Marandu.Toledo produced very few inflorescences in Years 1 and 3, and no inflorescences at all in Year 2.
Racemes per inflorescence declined with age, averaging 4.7 in the first year, 3.8 in the second year and 3.4 in the third year (Table 5).Overall, the majority of the BR06 lines produced fewer racemes per inflorescence than Mulato II.Five BR06 lines (0204, 0584, 1132, 1348 and 1696) produced numbers of racemes similar to or higher than Mulato II at each harvest.
Spikelet numbers per raceme were similar in the first and second seed harvests, 38 and 39, respectively, but declined to 34.6 at the third seed harvest (Table 5).Several BR06 lines produced more than 40 spikelets per raceme at each seed harvest, significantly higher than Mulato II and most BR02 lines.
BR02/0465 produced significantly heavier seed than all other lines and cultivars at all harvests, except for Toledo at the first harvest (Table 5).BR02/1794 and BR02/1718 produced significantly heavier seed than Mulato II and all 3 produced significantly heavier seed than all BR06 lines, except for BR06/0531, at the second and third harvests (Table 5).

Discussion
The main aims of the experiments were to identify hybrid brachiaria lines with seed yields higher than Mulato II and equal to or better than commercial seed yields of over 600 kg/ha produced by commercial brachiaria species in Australia and Brazil (Hopkinson and Clifford 1993;Souza 1999) 3).BR02/1794 usually flowers earliest (late September-early October).Peak flowering of BR02/1718 and BR02/0465 is usually 2 to 3 weeks later in mid-October, while peak flowering of Mulato II is nearly always in the second week of November in Northeast Thailand.
Different flowering times can strongly influence seed production.If late October-early November is particularly dry, late-flowering species can fail to set seed on sandy soils with low soil moisture retention.This appeared to be the case with Mulato II in the third year at Amnart Charoen (2010).Heavy rainfall in the first half of October benefited the earlier-flowering lines, but with no rain falling from late October onwards, Mulato II produced only 19 inflorescences per m 2 and only 19 kg/ha of seed was harvested.
The seed yields from the BR06 lines overall were disappointing.It was only when Mulato II failed to produce a large number of inflorescences at the third seed harvest at Amnart Charoen, that the BR06 lines produced more seed than Mulato II.However, these thirdharvest seed yields were also extremely low, averaging only 60 kg/ha.
Our experience with seed production of hybrid brachiaria grasses in Thailand is that seed yields decline with age, even though adequate levels of soil N are maintained by applying fertilizer.At the Ubon Ratchathani site seed yields from the first and second seed harvests were similar but at Amnart Charoen, seed yields were higher at the second harvest than at the first and very low for nearly all lines at the third harvest.The decline in seed yield over years in many tropical grass species is considered to be caused by larger tillers in older stands providing nutritional support for weaker tillers (low-yielding or sterile) to the detriment of their own seed development and the long-term productivity of the stand (Loch et al. 1999), though for the brachiaria lines we have no data to support this hypothesis.Farmers in Thailand have found that seed yields from secondyear hybrid brachiaria grass seed crops were less than half those of first-year seed crops.In order to get satisfactory seed yields, (300−400 kg/ha of clean seed), they treat hybrid brachiaria grass seed crops as annuals and replant every year, as they do with all other tropical grass seed crops (Hare 2014).
We consider that cleaned seed yields from commercial operations must be above 600 kg/ha for the seed prices of the hybrid brachiaria cultivars to become competitive with other commercial cultivars of brachiaria species.Some farmers in Thailand can produce more than 600 kg/ha of Mulato II seed by ground-sweeping but the majority produce only about 385 kg/ha (Hare 2014).Farmers in Northern Laos currently average 250 kg/ha of Mulato II seed from hand-knocking seed from seedheads (Hare 2014).In our experiments, we have at times produced 500 kg/ha of clean seed (98-99% purity by weight) (Hare et al. 2007b) by catching the seed in bags tied over the seedheads, but these occasions have been extremely rare.Commercial seed production of Mulato II is still very erratic.
Another factor which adds to the cost of hybrid brachiaria seed production is acid-scarification.This results in a loss of seed weight of 15-20% from scarifying off the glume, lemma and palea around each seed, light and empty seed, and small amounts of viable seed.Even though some viable seed is lost, without acidscarification, germination of the seed fails to exceed 30% (Hare 2014).
Nearly all hybrid lines produced sufficient numbers of inflorescences, racemes and spikelets to indicate a potential for useful seed yields.In the trials at Ubon Ratchathani, most hybrid lines produced 300-500 inflorescences/m 2 and at Amnart Charoen, the BR06 lines produced 300-500 inflorescences/m 2 in the first and second years.Mulato II produced fewer inflorescences than the new lines at both sites.Inflorescence numbers have nearly always been the main indicator of whether a forage plant has the potential to produce seed.However, with hybrid brachiaria grasses, it appears that seed-set is the most determining factor of seed yields.By seed harvest there seems to be a massive failure of seed-set, caryopsis maturation or both, with the cleaned seed coming from fewer than 10% of spikelets.In other brachiaria species it is not uncommon for abscission to precede maturation in a high proportion of spikelets (Hopkinson et al. 1996), but in the hybrid brachiaria grasses it appears to be an extremely high proportion.
Previous studies have shown seed yields of Mulato II are generally very low, with fewer than 2% of the spikelets formed producing viable seed (Hare et al. 2007a).There is speculation that this low seed-set is caused by pollen sterility, as Risso-Pascotto et al. (2005) found that more than 65% of pollen grains in brachiaria interspecific hybrids (B.ruziziensis x B. brizantha) were sterile and this sterility was genetic.Miles and Hare (2007) suggested this poor seed-set may be a common defect of newly formed apomictic forage grass hybrids.They referred to failures of buffel grass hybrids, which produced erratic and usually poor seed yields, leading to high seed prices.A hybrid-derived apomictic bahiagrass clone was not released because of concerns of low seed yields (Miles and Hare 2007).
In these current studies, we did not examine the percentage of spikelets that formed a caryopsis to calculate biological seed-set, but rather calculated economical seed-set, which is the ratio between realized and potential seed yield (Elgersma 1985).Mulato II seed-set (percentage of the number of cleaned seed to formed spikelets) ranged from 1.6 to 3.8% at Ubon Ratchathani and from 11.8 to 12.9% at Amnart Charoen.BR02/1794, which had superior seed yields to Mulato II, averaged 4.3% seed-set across both sites and seed-set for BR02/1752 ranged from 1.1 to 3.6% across both sites.Values for BR02/1718 were 1.8-3.1% at Ubon Ratchathani, and 7.0-12.2%at Amnart Charoen.Similarly, BR02/0465 had seed-set of 1.6-3.8% at Ubon Ratchathani, rising to 9.3-10.8%at Amnart Charoen.
The superior seed-set at Amnart Charoen compared with Ubon Ratchathani is interesting.Amnart Charoen is farther north than Ubon Ratchathani (15.5° vs. 15° N) and at a slightly higher elevation (168 vs. 130 masl).Grof (1968) showed that Basilisk signalgrass could set good seed yields and these seed yields were enhanced in drier upland regions in tropical latitudes (Loch et al. 1999).Basilisk seed production in Australia is predominantly on the Atherton Tablelands at lower latitudes but at elevations of 600-900 masl.In Brazil, successful seed production of Basilisk signalgrass and cv.Marandu is in the higher tropical latitudes (20 and 22° S) and at elevations of 700-1,000 masl (Souza 1999).The slightly higher elevation at Amnart Charoen compared with Ubon Ratchathani may have compensated for insuffi-cient latitude and encouraged greater seed-set.Ferguson et al. (1983) showed that, at similar latitudes in South America (15-19° S), the site with the highest elevation (1,000 masl) produced the highest seed yields of signalgrass, even though it had the lowest latitude (15° S).
Under commercial conditions in Thailand, we have produced hybrid brachiaria seed in Ubon Ratchathani, Amnart Charoen, Mukdahan and Roi-et provinces.It is only in the more northerly province, 160 masl), that farmers still continue with Mulato II seed production (Hare 2014).In the other provinces seed yields are too low and erratic to be economical and farmers have ceased production.Roi-et farmers, however, found that BR02/1752 seed yields (100-200 kg/ha) were too low to interest them; thus BR02/1752 seed production is limited to Northern Laos (19-21° N; 700-1,200 masl), where farmers find the seed yields satisfactory (200-300 kg/ha) under their low-input management (Hare 2014).We have also commenced seed production of BR02/1794 in these Northern Laos provinces.
While 43 hybrid Brachiaria lines were evaluated for seed yield between 2005 and 2010, only 3 lines, BR02/ 1794, BR02/1718 and BR02/0465, displayed a potential for seed yields greater than or equal to Mulato II.BR02/1752 had seed yields similar to or slightly lower than Mulato II, though in another study at Ubon Ratchathani, BR02/1752 and BR02/1794 produced significantly higher seed yields than Mulato II (Bouathong et al. 2011).
In considering the commercial release of hybrid brachiaria lines as named cultivars, forage production and quality (Hare et al. 2015) and seed production were important considerations, together with the waterlogging tolerance of BR02/1752, released as cv.Cayman (Pizarro et al. 2013), and the upright nature of BR02/1794 for cut-and-carry forage, released as cv.Cobra (E.Stern pers.comm.).
Further research is needed to verify the influence of elevation and latitude on flowering and seed-set in hybrid brachiaria grasses.In future breeding of new hybrids, there must be more rigorous selection for seed production characteristics at latitudes and sites typical of where commercial brachiaria seed production occurs.

Table 2 .
Field crop management of hybrid brachiaria lines during evaluation in Amnart Charoen, Thailand (Experiment 2).

Table 4 .
Seed yield and components of seed yield at peak anthesis of hybrid brachiaria lines during 2006 and 2007 in Ubon Ratchathani, Thailand (Experiment 1).

Table 5 .
Seed yield and components of seed yield at peak anthesis of hybrid brachiaria lines during 2008-2010 in Amnart Charoen, Thailand (Experiment 2).
. These aims were partially achieved.