PhenotyPIc varIatIon of Lotus cornicuLatus L . naturaL PoPuLatIons In reLatIon to habItat tyPe and management regImes

Lotus corniculatus is a highly diverse legume with good adaptability to different soil and climatic conditions. The objective of this research was to estimate the phenotypic diversity of individual L. corniculatus plants growing under two different habitat types with different grazing management. The collection of plant material was conducted in four forested and four open grassland sites in northern Greece. Plants from open grassland had significantly lower leaf area, leaf perimeter, leaf length, leaf width and internode length compared to those from the forest. This morphological differentiation could be contributed to differences in grazing intensity. Plants from open grasslands had significantly higher total phenols and condensed tannins, but lower crude protein content compared to those from the forest. Environmental factors in association with grazing management are the main contributors to the highly divergent concentrations of phenols, tannins and condensed tannins.


IntroductIon
Legume-based systems of forage production are economically attractive to dairy farmers in Europe and they confer an economic advantage to organic systems (Doyle and Topp, 2004).Additionally, they reduce the environmental impact of livestock farming (Rochon et al., 2004).The use of legumes is going to reinforce livestock farming as the political decisions for the Common Agriculture Policy in the EU are changed from high-input systems based on inorganic-N fertilizer to the low-input systems (Rochon et al., 2004).
Lotus corniculatus is a widely distributed legume of high nutritive value (Escaray et al., 2012), similar or even higher than that of Medicago spp.and Trifolium spp.It is characterized by good adaptability to different soil and climatic conditions, a fact that results in its genetic diversity (Steiner and Garcia de los Santos, 2001).The trait that differentiates L. cor-niculatus from the traditional forage legumes such as Medicago spp.and Trifolium spp. is its ability to accumulate phenolic compounds, especially proanthocyanidins, also known as condensed tannins (CT), in leaves (Cassida et al., 2000).The moderate concentration of CT (less than 50 g kg -1 DM) in forages reduces protein degradation (Frutos et al., 2004;Piluzza et al., 2014), and leads to lower ammonia production in the rumen and methane gas emissions (Hess et al., 2006).Moreover, this concentration increases the quantity of protein digested, prevents rumen bloat and controls internal parasite infections the use of anthelmintic drugs (Aerts et al., 1999;Min and Hart, 2003).On the contrary, a higher concentration of CT (more than 50 g kg -1 DM) affects negatively the nutritive value of forages, including digestibility, decreasing their palatability (Dixon et al., 2005) and reducing feed intake (Barry, 1989).The CT concentration in plant tissues is affected by genetic and environmental variables
Germplasm collections from natural populations of agriculturally important plant species have been widely used for the development of improved varieties.The diversity of natural L. corniculatus populations from different geographic regions has been studied using morphological traits (Steiner and Garcia de los Santos, 2001;Smith et al., 2009) and molecular markers such as AFLP and SSR (Savo Sardaro et al., 2008) and RAPDS (Steiner and Garcia de los Santos, 2001).These studies reveal evidence of ecotypic adaptation.Beside these studies, there is limited information about the diversity of natural populations of L. corniculatus regarding the content of phenolic compounds in relation to the growing environmental conditions.

materIaLs and methods
The study was conducted in the area of Taxiarchis on Mount Cholomondas, Chalkidiki prefecture, northern Greece (40°23'N, 23°28'E) at 800 m a.s.l.The climate of the area is classified as subhumid Mediterranean, with a mean air temperature of 11.1°C and an annual rainfall of 767 mm.Soils are mostly an association of Eutrophic Cambisols according to the FAO (1989) system developed on metamorphic rocks.The soils are acidic of mainly sandy loam texture.The area is situated in the Quercion confertae subzone of the Quercetalia pubescentis (sub-Mediterranean) zone (Athanasiadis, 1986).The entire forested area, around 2500 ha, is public and is communally grazed by sheep and goats.L. corniculatus is present in the area mainly in the open grasslands and at the edge of forest.The collection of plant material was conducted in four forested and four open grassland sites.Grazing intensity was varied from non-grazed to lightly grazed in the forested sites and from non-grazed to heavily grazed in the open grasslands (Table 1).Twenty individual plants at the flowering stage were randomly collected from each of the forested and open grassland sites in August 2011, i.e. a total of 160 plants.The collected plants were visually recorded in situ for growth habit (1 = prostrate and 2 = erect) and flower color (on a scale of 1 − light yellow to 4 − red-yellow) before their above ground biomass was cut and transferred to the lab.
Three branches of each individual plant were scored for the following morphological traits: leaf area (cm 2 ), leaf perimeter (cm), leaf length (cm), leaf width (cm), and internodes length (cm).The three branches of each plant were placed on an A4 white page and scanned.The image processing and the measurements of the morphological traits were performed using the software Image-Pro Plus 6.3 (Media Cybernetics).
All the samples were oven-dried at 50 o C for 48 h, ground through a 1-mm sieve and analyzed for N using the Kjeldahl procedure (AOAC, 1990).Crude protein (CP) concentration was then calculated by multiplying the N content by 6.25.Samples were analyzed for total phenols (TPH), total tannins (TT) and condensed tannins (CT) according to Makkar (2003).Three replicates of 200-mg samples were extracted in 10 ml of aqueous acetone (acetone: water, 7:3) twice in an ultrasonic water bath for 20 min.The extracted samples were centrifuged at 3000 g at 4°C for 10 min and the supernatants were used for tannin analysis on the same day (Makkar, 2003).TPH and TT in the extract were determined by a modification of the Folin-Ciocalteu method using polyvinylpolypyrrolidone (PVPP) to separate tannin phenols from non-tannin phenols (Makkar et al., 1993).The concentration of TT (mg/g DM) was calculated as follows: TT conc.= (conc.of TPH) -(conc. of TPH remaining after PVP treatment).Both TPHs and TTs were expressed as tannic acid equivalent (mg/g TAE).
Condensed tannins were determined according to the method of Porter et al. (1986), using purified Quebracho CT as the reference standard.The CT contents are therefore expressed as Quebracho equivalents.
The means, range and coefficient of variation (CV) of all phenotypic traits (morphological and chemical) were calculated for each habitat (forest vs. grassland) and for each collection site in each habitat.All phenotypic traits were tested for normality using the Kolmogorov-Smirnov test.Log data transformations were performed for the non-normally distributed traits.One-way ANOVA of the normally distributed data and Kruskal-Wallis ANOVA of the non-normally distributed data were performed using SPSS® statistical software v. 18.0 (SPSS Inc., Chicago, IL, USA), in order to determine differences among the the habitats and the collection sites in each habitat.The LSD for the normally distributed and Mann-Whitney test for the non-normally distributed traits at the 0.05 probability level was used to detect the differences among means (Steel and Torrie, 1980).
The data of the morphological traits were also used to generate eigenvalues, the percentage of variation accumulated by PCA, and the load coefficient values between the original characteristics and respective PCA.The first two principal components, which accounted for the highest variation, were used to plot the 2-dimensional dispersion of the collection sites.

resuLts
Leaf perimeter, leaf length, leaf width, TPH and CT content were the phenotypic traits that were normally distributed.Leaf area was normally distributed after log transformation.Inversely, growth habit, flower color, internode length, TT and CP content remained nonnormally distributed even after log transformation.
Significant variation among the habitats was detected for all the normally and non-normally distributed morphological traits except the growth habit (Table 2).In particular, the L. corniculatus plants from open grassland had significantly lower leaf area, leaf perimeter, leaf length, leaf width, and internodes length, while they had darker flower color compared to those from the forested sites (Table 2).
The morphological traits significantly differed among the grassland sites, while they did not differ among the forested ones (Table 3).The heavily grazed site of 'Bou' had the smallest leaves and internode length, the most prostrate growth and the most redyellow flowers (Table 3).
Additionally, the morphological traits of plants from the grasslands generally revealed higher variability in terms of CV estimation compared to those from the forested areas (Table 2).Among the grasslands and forested sites, the plants from the moderately grazed sites tended to have higher variability in terms of CV compared to those from the non-grazed and heavily grazed ones (Table 3).
The first three principal components (PCs) with Eigen values >1.0 cumulatively contributed about 94% of the total variation among the accessions for the seven morphological traits (Table 4).PC1, PC2 and PC3 accounted for 53%, 21% and 20%, respectively, of the total variation.The variation in PC1 was mainly associated with leaf characteristics and particularly with leaf area and leaf width.The predominant traits in PC2 were growth habit and in PC3 internode length.According to the scatter diagram of the first two principal components (Fig. 1), all the grassland sites were located on the negative side of PC1, except the non-grazed grassland site 'Liv' , which was located on the positive side together with all the forested sites.
Similar to the morphological traits, significant variation among the habitats was detected for all the normally and non-normally distributed chemical traits except for the TT content (Table 5).The plants from open grasslands had significantly higher TPH and CT, but lower CP content compared to those from the forested sites.
Among the forested sites, L. corniculatus plants from the non-grazed areas of 'Dou' and 'Fyt' had significantly lower TPH, TT and CT, but significantly higher CP content compared to plants from the lightly grazed areas (Table 6).Similarly, plants from the non-grazed grassland site of 'Liv' had significantly lower TPH and TT content, while the CT and CP did not significantly differ among them (Table 6).In contrast to the morphological traits, the plants from forested sites had higher CV values in comparison to those from grasslands for all the chemical traits.

dIscussIon
Differentiation in the morphological and chemical traits of the L. corniculatus plants was observed among the habitat types.Regarding the morphological traits, the plants from the open grassland were characterized by smaller plant parts and darker flowers compared to those from the forest.Similarly, morphological differentiation was recorded among habitat types within a region in British collections of L. corniculatus (Smith et al., 2009).This morphological differentiation could be attributed either to environmental selection pressures at each habitat or to different management regimes (Peter-Schmid et al., 2008).
The examined open grassland sites with high variation in grazing intensity were morphologically differentiated while this was not the case with the forested ones.This result suggests that the morphological differentiation between the habitats could be mainly attributed to grazing intensity.This suggestion is further supported by the PC results, according to which the orientation of the tested sites depended on grazing intensity.According to Diaz et al. (1992), the increase in grazing intensity has resulted in progressive miniaturization of the plants' photosynthetic structure and close to the ground biomass concentration.Generally, smaller and more prostrate plants of heavily grazed populations compared to that of moderately or non-grazed ones have been reported for many plant species, such as the annual forb Persicaria longiseta (Suzuki, 2008), the forage grass Sesleria albicans (Reisch and Poschlod, 2003) and other forage grasses (Painter et al., 1993).This small and prostrate morphology is the response of plants in order to reduce damage by grazing animals.Morphological variation tends to be higher on the open grassland sites and moderately grazed ones compared to the forested and non-grazed or heavily grazed ones, respectively.Reduced plant population variability under heavy grazing is expected due to strong directional selection (Falconer and Mackay, 1996).However, contrasting results have been reported about the effect of grazing intensity on a population's variability.For instant, Völler et al. ( 2013) have reported that the within population variability of B. hordeaceus increased with the increase in grazing intensity.
Regarding the chemical traits, the plants from open grasslands were characterized by a higher content of phenolic compounds and lower content of CP.The CT content was generally less than 50 g kg -1 DM, indicating a positive effect on plant nutritive value (Piluzza et al., 2014), although some individuals from the grasslands were very close to this threshold (Table 5).Environmental factors such as light intensity, water stress and plant phenological stage could be attributed to this differentiation (Van Soest, 1994).According to Iason and Hester (1993) and Jantineke et al. (2009), low light intensity reduced the TPH, TT and CT content in plants.On the other hand, it is well known that shade improves the nutritive value of herbaceous vegetation as the maturity stage is delayed compared to open field conditions (Burner andMacKown, 2006, Parissi andKoukoura, 2009).The increase of CP content in legume species under shade has also been reported by Kyriazopoulos et al. (2013).
Similar to the morphological traits, it seems that the chemical traits are also affected by grazing intensity.
Plants from non-grazed areas of both open grassland and forested sites tended to have a higher content of phenolic compounds.Contrary to the morphological traits, a higher variation in chemical traits was observed in the forested sites compared to the open grassland ones.It is well documented that plant-herbivore interactions lead to the development of defense mechanisms of plants against herbivores, by producing tannins (Barroso et al., 2001).Furthermore, the observed variability in flower color among the studied populations probably suggests differentiation in the composition of phenolic compounds.According to Jay and Ibrahim (1986), the yellow color of the L. corniculatus flower is dependent on the composition of flavonoids.

concLusIons
The morphological differentiation of the studied populations of L. corniculatus could be mainly at-tributed to grazing intensity.Environmental factors in association with grazing management are the main contributors to the highly divergent concentrations of phenols, tannins and condensed tannins.Further studies are needed in order to better understand how these factors affect the concentration of phenols in L. corniculatus germplasm.
authors' contribution: Panagiota Giagourta: field and laboratory measurements; Eleni M. Abraham: experimental design, statistical analysis, manuscript writing; Zoi M. Parissi: laboratory chemical analysis, manuscript writing; Apostolos P. Kyriazopoulos: experimental design, manuscript writing references Aerts, R.J., Barry, T.N. and W.C McNabb (1999) The objective of this research was to estimate the phenotypic diversity of individual L. corniculatus plants growing under two different habitat types, namely in forest and in open grassland, with different grazing management regimes.

table 1 .
Geographic locality and ecological description of collection sites of natural populations of Lotus corniculatus on Mount Cholomondas, northern Greece (Gr).

geographic Localities/sites code geographic data altitude aspect dominant herbaceous species description/ management forest Latitude Longitude
*Source: Taxiarchis Forest Services Management Plan

table 2 .
Means (M), range (R) and coefficient of variation (CV) of morphological traits of Lotus corniculatus from forest and grassland sites.

table 3 .
Means (M), range (R) and coefficient of variation (CV) from the four different forest and grassland sites.
*Means followed by the same letter in the same row do not significantly differ P≥0.05

table 4 .
Percentage, cumulative variances and Eigen vectors for the eight principal components of each morphological trait of Lotus corniculatus.
fig. 1. Scatter plot of the first and second principal components based on the morphological traits of Lotus corniculatus accessions from the edge of forest () and grassland ().

table 6 .
TPH (mg/g DM TAE), TT (mg/g DM TAE), CT (mg/g DM QE) and CP (%) concentration of individual Lotus corniculatus plants from the collecting sites in forest and grassland.
*Means in the same row within each vegetation type followed by the same letter do not significantly differ (P≥0.05)