Preliminary Results on Agronomic Performance of Barrel Medic ( Medicago truncatula ) in Serbia

A small-plot trial with eight Australian barrel medic (Medicago truncatula Gaertn.) cultivars was carried out in 2010 at Rimski Šančevi. The average green forage and forage dry matter yields were highest in Jemalong (30.7 t ha-1 and 7.3 t ha-1) and Parabinga (30.7 t ha-1 and 8.0 t ha-1). Forage dry matter crude protein content ranged between 150.8 g kg-1 in Parabinga and 179.4 g kg-1 in Jester. Forage neutral detergent fi bre content varied from 305.2 g kg-1 in Sephi to 458.8 g kg-1 in Caliph, while the average forage acid detergent fi bre content was 312.8 g kg-1. The average seed yield for all cultivars was 281 kg ha-1 and may be considered satisfying, as it was obtained in a very rainy and warm growing season. Jemalong and Parabinga had the highest aboveground nitrogen yield (190 kg ha-1 and 193 kg ha-1) and thus the greatest potential for green manure.


Introduction
The most important forage fi eld legumes in Serbia are lucerne (Medicago sativa L.), red clover (Trifolium pratense L.), pea (Pisum sativum L.) and vetches (Vicia spp.).It is estimated that the cultivation area of lucerne varies between 180,000 ha and 200,000 ha, while red clover takes from 80,000 ha to 120,000 ha (Karagić et al. 2010).Lucerne is dominant in the northern parts of the country with rich chernozem soils, while red clover is more widely distributed in central Serbia, being more tolerant to prevailing acid soils.Pea and vetches are the most signifi cant annual forage legumes, appreciated for the diversity of their use, ranging from green forage, forage dry matter and forage meal, over silage and haylage to dry grain and straw (Mikić et al. 2010a).Field pea (comprising forage and dry pea) is grown on about 25,000 ha in Serbia today, while vetches are grown on 7,500 ha (Mihailović et al. 2005).
Barrel medic (Medicago truncatula Gaertn.)originated in the Mediterranean basin (Rubiales et al. 2009) and became widely naturalised in other regions of the world such as Australia, where annual medics are used as forages and soil improvers on about 50 million ha (Crawford et al. 1989).During the last decade of the last century, barrel medic was recognised as appropriate to be a model species for legumes, especially due to its diploid and small (about 5 10 8 base pairs) genome, its self-fertile nature, its prolifi c seed production and its rapid generation time (Cook 1999, Young et al. 2003, Tivoli et al. 2008, Thompson et al. 2009).This was confi rmed by various researches revealing a high synteny, i.e. an extensive conservation of gene order, not only between barrel medic and lucerne (Choi et al. 2004), but also between barrel medic and pea (Kaló et al. 2004).In this way, efforts in genomic research of barrel medic give new perspectives for lucerne breeding (Annicchiarico & Scotti 2009, Julier & Meusnier 2010).The oldest and largest collection of barrel medic is housed at the South Australian Research and Development Institute (SARDI), with more than 5,000 accessions (Ellwood 2006).
The main goal of this very preliminary research was to assess the possibility of growing barrel medic for forage production in the prevailing agro-ecological conditions of northern Serbia.It was especially aimed at assessing the length of its growing season when cultivated for forage and thus the possibility to include it as a shortseason crop in prevailing cropping systems in Serbia, with potentially preceding autumn-sown brassicas and succeeding sorghum, Sudan grass or short-season maize.At the same time, this research was aimed at determining the potential of barrel medic for seed production in the conditions of Serbia, as well as its hypothetical use as green manure crop.The trial was established on April 23.All eight cultivars were sown at a density of about 250 viable seeds m -2 (Walsh et al. 2001), a plot size of 1 m 2 , three replicates for forage production and one plot for seed production, a row spacing of 20 cm both within and between plots in order to reduce excessive growth of border rows of a plot.In fact, each cultivar was sown in seven rows, but only fi ve inside the plot were taken into account, while two border rows were discarded.The dominant weather and soil conditions during the trial period at Rimski Šančevi in 2010 are given in Tables 1 and 2.

A
All eight cultivars were cut in full bloom and while forming fi rst pods, being a desirable balance between yield and quality in majority of annual forage legumes (Mihailović et al. 2008).Quantitative morphological characteristics, such as average stem length (cm), number of stems (plant -1 ), number of leaves (plant -1 ), number of fl owers (plant -1 ) and number of leaves (plant -1 ) were measured on the plant samples taken before the cutting.The plant samples of 30 plants per replication were taken from the middle rows of each plot.The same samples were used to measure the proportion of each plant part, namely stems, leaves, fl owers and pods, in the total green forage yield and, after they were dried at the room temperature to a constant mass, in the total dry forage matter yield.Green forage yield (t ha -1 ) was calculated on the basis of the green forage yield (kg m -2 ), measured immediately after the cutting, while forage dry matter yield (t ha -1 ) was calculated from plot fresh yield and forage moisture of the forage samples.
The chemical analysis of the forage dry matter in all eight cultivars included crude protein content (g kg -1 ), by multiplying the forage dry matter nitrogen content (g kg -1 ) by 6.25, and the content (g kg -1 ) of neutral detergent fi bre (NDF), acid detergent fi bre (ADF) and lignin.Forage crude protein yield (kg ha -1 ) was calculated on the basis of the forage dry matter crude protein content and forage dry matter yield.
In the stage of full maturity of the seeds in the fi rst pods, all eight cultivars were harvested by hand.Seed yield (kg ha -1 ) was calculated on the basis of the seed yield (g m -2 ), at a moisture content of 14%.Thousand seed weight (g) and seed viability (%) were determined by the samples of 100 seeds one month after the harvest of each cultivar.In order to enable a basic statistical analysis with one plot only, the separate samples were taken from three middle rows, with the seed yield of each recalculated for the statistical purposes.
On the basis of forage dry matter nitrogen content and forage dry matter yield, aboveground nitrogen yield (kg ha -1 ) was calculated in order to assess the potential of the tested barrel medic cultivars for green manure.
Number of days from sowing to cutting and number of days from sowing to harvest for each of eight cultivars were also recorded.
The results were processed by the method of analysis of variance (ANOVA) with the LSD test, using the MSTAT-C software.

Results and Discussion
Weather and soil conditions The growing season of barrel medic at Rimski Šančevi was by 1°C warmer in comparison to long-term average (Tab.1).At the same time, fi ve months of the barrel medic growing season at Rimski Šančevi were characterised by almost twice precipitation relative to long-term average.
The chernozem soil at Rimski Šančevi was well provided with basic mineral nutrients and humus, with a slight alkaline reaction (Tab.2), and thus might be considered favourable for the barrel medic cultivation.

Forage yield
There were signifi cant differences among the average values of number of days from sowing to cutting and quantitative morphological characteristics related to forage yield in eight tested barrel medic cultivars (Tab.3).Number of days from sowing to cutting varied between 55 in Parabinga and 71 in Mogul, pointing that barrel medic has much shorter growing season for forage production in comparison to traditional annual forage legumes such as common vetch, where it varies between 88 days and 102 days (Mikić 2008).On average, stem length in barrel medic was 40 cm, similar to plant height in the fi rst cut of lucerne, especially in the year of establishment (Mikić et al. 2007).With the number of stems ranging from 2.2 plant -1 in Paraggio to 7.1 plant -1 in Mogul, the barrel medic cultivars, in average, were moderately prone to basal branching in comparison to other annual forage legumes such as largefl owered (Vicia grandifl ora Scop.) and other vetches (Mikić et al. 2009).The highest number of leaves was in Parabinga (136 plant -1 ).The cultivars Caliph and Parabinga had greater number of pods (32 plant -1 and 38 plant -1 ) than fl owers in cutting and as compared to other six cultivars, witnessing their earliness and a faster transition from fl owering to pod forming.The average green forage and forage dry matter yields were highest in Jemalong (30.7 t ha -1 and 7.3 t ha -1 respectively), and Parabinga (30.7 t ha -1 and 8.0 t ha -1 respectively) (Tab.4).The forage dry matter yield in most of the tested cultivars at Rimski Šančevi was within the limits barrel medic achieved in the upper Midwest of USA on Hubbard loamy sand, ranging from about 4.7 t ha -1 to 9.1 t ha -1 (Zhu et al. 1998).This confi rms that barrel medic has the potential for producing high forage yields in temperate regions such as the north-central USA and the northern Balkans, as well as that it might be used when traditional forage supplies are inadequate (Zhu et al. 1996).
It is considered that a higher ratio between leaves and stems in lucerne is mainly associated with shorter and thinner main stems, as well as that it is modestly and inversely correlated with forage yield (Annicchiarico 2007).In all eight  (49 : 42).Caliph and Parabinga were the only cultivars where the joint proportion of fl owers and pods was higher than 10%.

Forage quality
Barrel medic is considered a valuable source of forage protein, especially in mixtures with grasses (Stout et al. 1997).The average forage dry matter crude protein content in eight tested barrel medic cultivars ranged between 150.8 g kg -1 in Parabinga and 179.4 g kg -1 in Jester (Tab.6).This was lower in comparison to the values of forage dry matter crude protein content in lucerne cultivars, with a variation from 177 g kg -1 to 194 g kg -1 (Katić et al. 2009), and could partially be explained by an excessively rainy barrel medic growing season in 2010.Paraggio and Sephi had signifi cantly lower forage crude protein yield (511 kg ha -1 and 473 kg ha -1 ) than the rest of the tested barrel medic cultivars, led by Parabinga (1,206 kg ha -1 ).Due to the prevailing conditions during its growing period in 2010, barrel medic had a lower average forage crude protein yields in comparison to other annual forage legumes such as pea, with more than 1,500 kg ha -1 (Mihailović et al. 2010).
The content of forage NDF in eight barrel medic cultivars varied between 305.2 g kg -1 in Sephi and 458.8 g kg -1 in Caliph and with an average value of 363.5 g kg -1 .In this way, barrel medic had lower forage NDF content than lucerne, with an average value of 391 g kg -1 (Julier et al. 2000) and almost equal to an average seasonal value of 369.3 g kg -1 in red clover (Vasiljević et al. 2008).In a similar way, the average forage ADF content in eight barrel medic cultivars (312.8 g kg -1 ) was lower in comparison to that in lucerne cultivars in the same agro-ecological conditions, with a range from 374.4 g kg -1 to 395.2 g kg -1 (Katić et al. 2008), but also higher than an average seasonal value of 289.1 g kg -1 in red clover (Vasiljević et al. 2009).Sephi had the lowest forage lignin content (62.5 g kg -1 ), while had the highest forage lignin content (103.8 g kg -1 ).

Seed yield
The number of days from sowing to harvest in eight barrel medic cultivars varied between 90 in Jester and Parabinga and 106 in Mogul (Tab.7), with a similar general trend as the number of days from sowing to cutting and confi rming once again that barrel medic also has shorter growing season for seed production in comparison to the traditional Serbian annual forage legumes such as common vetch, where it varies between 116 days and 136 days (Mikić 2008).The average seed yield in the tested barrel medic cultivars was 281 kg ha -1 and may be considered satisfying bearing in Table 6.Parameters related to forage dry matter chemical composition (g kg -1 ) and forage crude protein yield (kg ha -1 ) in barrel medic cultivars at Rimski Šančevi in 2010 Tabela 6. Veličine vezane za hemijski sastav suve materije krme (g kg -1 ) i prinos sirovih proteina krme (kg ha  (Gallardo et al. 2003).

Conclusions
Although preliminary, the results of this agronomic performance evaluation of barrel medic in Serbia show that it has a certain potential for forage production.It is noteworthy that, even in a rainy and a warm spring and summer of 2010, it achieved satisfying forage yields of good quality and, in several cases, reliable seed yields, both during a relatively brief growing season.In that way, barrel medic could fi nd its place as a short-lived forage or green manure crop in diverse cropping systems with benefi ts for both animal husbandry and soil fertility.At the same time, its promising preliminary agronomic performance opens a possibility for the evaluation of the results of various genomic research in fi eld trials on chernozem and other prevailing soil and weather conditions in Serbia.Another fi eld trial season with barrel medic has already begun with establishing the evaluation of cold tolerance in the autumn 2010.
small-plot trial was carried out in 2010 at the experimental fi eld of the Institute of Field and Vegetable Crops at Rimski Šančevi, near Novi Sad, 45 o 20' N, 19 o 51' E and 84 m asl.It included eight barrel medic cultivars of Australian origin, namely Borung, Caliph, Jemalong, Jester, Mogul, Parabinga, Paraggio and Sephi, kindly provided by SARDI.

Table 2 .
Chemical properties of the surface chernozem soil layers during the trial with barrel medic at