The variation of the chemical composition of the main plant species in a subalpine grassland in northwestern Greece

This study presents the results of a survey conducted in the year 2013 in a subalpine grassland located in Epirus, in northwestern Greece, and concerns the determination of the chemical composition of nine wild plants species (Alopecurus gerardii V., Stipa pennata L., Phleum alpinum L., Trifolium repens L., Lathyrus aphaca L., Lotus corniculatus L., Ranunculus repens L., Achillea millefolium L., and Geranium lucidum L.) at various stages of growth. The highest mean neutral detergent fiber content was determined in S. pennata L. samples with 546 g kg−1 dry matter (DM). The highest mean acid detergent fiber content was determined in samples of A. gerardii L. with 358 g kg−1 DM. For calcium (Ca), the highest average content was determined in the samples of L. corniculatus L. with 9.6 g kg−1 DM. For phosphorus (P), iron (Fe), and copper (Cu), the highest average contents were determined in the samples of R. repens L. with 2.8 g kg−1 DM, 60.2 mg kg−1, and 5.6 mg kg−1 DM, respectively. For magnesium (Mg), the higher average content was found in L. aphaca L. samples with 1.6 g kg−1, for potassium (K) in G. lucidum L. samples with 15.6 g kg−1, for sodium (Na) in A. millefolium L. samples with 15.3 g kg−1, and for manganese (Mn) in P. alpinum L. samples with 37.9 mg kg−1. The chemical composition of plant species meets the nutritional requirements of grazing animals in the investigated area. Among the plant species, legumes and other forbs have the highest nutritional value.

Generally, the higher the crude fiber content of the plants, the lower their nutritional value for the animals (Papanastasis & Ispikoudis, 2013).
Plants differ in their chemical composition. These differences are evident both between plants and during their various stages of growth (Elgersma & Søegaard, 2016). The same plant species at the same stage of growth differ significantly in their chemical composition when grown in different environments, even when in nearby environments (Belkhir et al., 2013;Buxton, 1996;Tamburino et al., 2012).
When the mineral content of the grassland is low, the quality of the grassland production is affected as animal feed intake decreases, thereby adversely affecting microbial growth and activity of the large abdomen. This leads to a decrease in the digestibility of forage (Leng, 1990;Provenza, 1995). The forage content in minerals depends on many factors such as soil, climate, season of the year, plant species and plant growth, and fertilization (Flisch, Sinaj, Charles, & Richner, 2009;Georgievskki, Annenkov, & Samokhin, 1982;Öborn et al., 2008;Ramírez-Pérez, Buntinx, & Rosiles, 2000).
In grasslands, P is usually the most inadequate element. Its lack in livestock feed has resulted in reduced livestock reproduction (Hale & Olson, 2001;McDowell, Ellis, & Conrad, 1984;McDowell & Valle, 2000). In order to meet the dietary needs of ruminants, the rangeland plants should contain at least 3 g kg −1 Ca, 2 g kg −1 Mg, and 8 g kg −1 K (Tajeda, Mcdowell, Martin, & Conrad, 1985). According to Ward and Lardy (2005), the forage has a relatively high K content ranging from 10 to 20 g kg −1 dry matter (DM). Also, the copper (Cu) content of rangeland plants adequately meets the needs of most livestock (Corah & Dargatz, 1996;Greene, 2000;Greene, Baker, & Hardt, 1989).
The mineral content of plants varies depending on their stage of growth. In general, the mineral content of plants decreases as the plants mature (Pirhofer-Walzl et al., 2011;Schlegel, Wyss, Arrigo, & Hess, 2016). Tan, Temel, and Yolcu (2003) reported that the K, Mg, Ca, and P content of plants decreased from 29.31 to 22.04, 3.48 to 2.85, 12.83 to 11.58, and 1.50 to 1.19 g kg −1 , respectively, as the plants matured.
Sheep needs in Ca range from 2 to 8.2 g kg −1 DM, in P, the minimum requirements are 1.6 g kg −1 DM, in Mg range from 1.2 to 1.8 g kg −1 DM, in K from 5 to 8 g kg −1 DM, in Na from 0.9 to 1.8 g kg −1 of DM, in Fe, the minimum requirements are 30 mg kg −1 DM, in Mn are 20 mg kg −1 DM, and in Cu range from 20 to 30 mg kg −1 DM (National Research Council, 1985). According to Zervas (2013), the daily maintenance requirements of a 50-kg sheep are 5 g day −1 in Ca, 0.7 g day −1 in Mg, 3.5 g day −1 in P, and 1.25 g day −1 in Na, whereas the minimum requirements in Mn and Cu are 30 and 5 ppm day −1 , respectively.
The aim of the present study was to determine the variation in the chemical composition of the main plants of the subalpine grassland of "Kostilata" by botanical group, as their chemical composition determines the nutritional value of the forage. The knowledge of the nutritional value of plants will contribute to the rational management of grazing as well as to the preservation and protection of the ecosystem.

| Study area
The study was conducted in 2013, in "Kostilata" subalpine grassland.
This area extends at an altitude of 1,100 to 2,393 m and it is located in Theodoriana, Epirus, Western Greece. It occupies an area of 950 ha with 3,500 sheep grazing during the summer months.
In "Kostilata" grassland, there are 96 taxa. Within the grassland itself, there is a variation in the physicochemical properties of the soil and its inclinations, depending on the altitude. At an altitude of 1,100 to 1,400 m, the soil contents of sand, silt, clay, organic matter, available P, and pH value were found at 43.9%, 38.3%, 17.8%, 6.9%, 4.2 mg kg −1 , and 5.9, respectively. At an altitude of 1,401 to 1,800 m, the soil contents of sand, silt, clay, organic matter, available P, and pH value were found at 48.7%, 35.1%, 16.2%, 6.7%, 5.6 mg kg −1 , and 5.5, respectively. At an altitude of 1,801 to 2,393 m, the soil contents of sand, silt, clay, organic matter, available P, and pH value were found at 52.9%, 33.9%, 13.2%, 5.5%, 6.2 mg kg −1 , and 5.4, respectively. Most of the mild slopes (from 0% to 30%) are at an altitude of 1,100 to 1,400 m, whereas most of the steep slopes (over 60%) are at altitudes above 1,800 m (Roukos et al., 2016).

| Plant samplings analyses
Sixty (60) fixed experimental cages of 4 m × 4 m in size were installed and enclosed by a fence of 1 m high, in order to protect the plants from grazing. The cages were placed randomly and in such a way so as to be representative of the grassland. Samples were collected approximately every 10 days, from late April to mid-July. Three plant species were selected from each plant group (grasses, legumes, and other forbs) that contained the highest percentage of biomass.
For the identification of the plant species, the encyclopedia "Mountain Flora of Greece I and II" (Strid, 1986;Strid & Tan, 1991), the book "The main grasses of natural grasslands" (Papanastasis & Karagiannakidou -Papadimitriou K., 1993), and the book "Vascular Plants of Greece" (Dimopoulos et al., 2013) were used. The samples were placed in an oven for drying at 65 C for 48 hr (Deinum & Maassen, 1994) and milled to particle size of 1 mm using a Kinematica Polymyx PX-MFC 90 D cutting mill.
The NDF content was determined according to the method of Van Soest, Robertson, and Lewis (1991) as modified by Vogel, Pedersen, Masterson, and Toy (1999) using an ANKOM 2000 fiber analyzer. The samples were analyzed using heat stable amylase without sodium sulfite in the neutral detergent reagent. The ADF content was determined according to AOAC (1997, method 973.18). NDF and ADF were expressed without residual ash. P was determined spectrophotometrically using a HITACHI, U-1800 spectrophotometer. Ca, Mg, K, Na, Cu, Fe, and Mn were determined by atomic absorption spectrometry using a PERKIN ELMER atomic absorption spectrometry instrument (AOAC, 1999, method 968.08).
A weather station that was already installed at an altitude of 1,100 m was used to obtain the climatic parameters (air temperature and precipitation). For the same purpose, two new weather stations were installed, in spring 2013, at an altitude of 1,600 and 2,050 m, respectively.

| Statistical analysis
The results were compared for significant differences by one-way analysis of variance test, whereas mean differences were checked using Tuckey's test (P < .05). Statistical analyses were performed with OriginPro 9.0 software.

| Precipitation and temperature
Different climatic conditions prevail within the "Kostilata" grassland.
Higher temperatures at lower altitudes were recorded throughout the investigation. In May, higher precipitation was observed at lower altitudes, whereas in June and July at higher altitudes ( Table 1). The last 10 days of April, only R. repens L. species was found in the grassland, whereas P. alpinum L. species was found for the first time in the first 10 days of June.

| NDF and ADF content
The highest mean NDF content was found in the samples of S. pennata L. species with 546 g kg −1 DM, with a statistically significant difference (P < .05) from the other plant species except A. gerardii V. (Table 2). The highest mean ADF content was found in the samples of A. gerardii V. species with 358 g kg −1 DM, with statistically significant difference (P < .05) from other plant species, except S. pennata L. and A. millefolium L. (Table 3). In all plant species, the lowest NDF and ADF contents were determined in the samples of the first samplings and then a gradual increase in NDF and ADF values was observed, with statistically significant differences (P < .05). Also, statistically significant differences (P < .05) were observed between the values of different plant species at the same sampling dates (Tables 2 and 3).

| Macrominerals
Concerning the average mineral content of plant species, a significant variation was observed. The highest mean Ca content occurred in the samples of L. corniculatus L. species with 9.6 g kg −1 DM, with statistically significant differences (P < .05) observed only with the mean values of grasses. Between sampling times, statistically significant differences (P < .05) were observed only in grass samples (Table 4).
The highest mean P content was found in R. repens L. species with 2.8 g kg −1 DM, with statistically significant differences (P < .05) observed only with the mean values of legumes and S. pennata L. species. Statistically significant differences between sampling periods were observed only in T. repens L. and R. repens L. species (Table 5).
The highest average Mg content was found in L. aphaca L. species with 1.6 g kg −1 DM. Statistically significant differences (P < .05) were observed between the mean values of L. aphaca L. and the mean values of grasses and G. lucidum L. species (Table 6). For K, the highest mean content was found in the samples of G. lucidum L. species with 15.6 g kg −1 DM, without any statistically significant differences (P < .05) with the mean values of other plants. Statistically significant differences between sampling periods were observed in A. gerardii V., S. pennata L., L. aphaca L., R. repens L., A. millefolium L., and G. lucidum L. species (Table 7).
For Na, the highest average content was found in A. millefolium L. species with 15.3 g kg −1 DM, with statistically significant difference observed with mean values of grasses and L. corniculatus L. species. Among the sampling periods, no statistically significant differences were observed for T. repens L. and L. aphaca L. species only (Table 8).

| Microminerals
The highest average Fe content was found in R. repens L. species with 60.2 mg kg −1 DM, with statistically significant differences observed only between the above value and the mean values of S. pennata L., P. alpinum L., and G. lucidum L. species. Statistically significant differences between sampling periods were observed in T. repens L., L. corniculatus L., and A. millefolium L. species (Table 9).
The highest average Mn content was observed in P. alpinum L. species with 37.9 mg kg −1 DM, with statistically significant differences (P < .05) observed between the mean values of legumes and G. lucidum L species. Statistically significant differences between sampling periods were observed in A. gerardii V. and A. millefolium L. species (Table 10).
Concerning Cu, the highest average content was found in R.
repens L. species with 5.6 mg kg −1 DM, with statistically significant differences (P < .05) observed in all other plant species, except L. aphaca L. and A. millefolium L. Statistically significant differences between sampling periods were observed in all plant species, except L. aphaca L., L. corniculatus L., and A. millefolium L. (Table 11).
T A B L E 1 Precipitation and air temperature during the study period on the "Kostilata" grassland Note. Mean values followed by different letters (a, b, c, d, and e), in the same row, differ significantly (P < .05). Mean values followed by different exponents ( 1, 2, 3, and 4 ), in the same column, differ significantly (P < .05).
T A B L E 3 Acid detergent fiber content of plant species, at various stages of growth (g kg −1 dry matter) Note. Mean values followed by different letters (a, b, c, d, and e), in the same row, differ significantly (P < .05). Mean values followed by different exponents ( 1 , 2, 3, and 4 ), in the same column, differ significantly (P < .05).
T A B L E 4 Calcium (Ca) content of plant species, at various stages of growth (g kg −1 dry matter) Note. Mean values followed by different letters (a and b), in the same row, differ significantly (P < .05). Mean values followed by different exponents ( 1, 2, 3, and 4 ), in the same column, differ significantly (P < .05).
T A B L E 6 Magnesium (Mg) content of plant species, at various stages of growth (g kg −1 dry matter) Note. Mean values followed by different letters (a and b), in the same row, differ significantly (P < .05). Mean values followed by different exponents ( 1, 2, 3, 4, and 5 ), in the same column, differ significantly (P < .05).
T A B L E 7 Potassium (K) content of plant species, at various stages of growth (g kg −1 dry matter)  (a and b), in the same row, differ significantly (P < .05). Mean values followed by different exponents ( 1 and 2 ), in the same column, differ significantly (P < .05).

| DISCUSSION
The higher content of plant species of the botanical group of grasses in NDF and ADF, with statistically significant differences (P < .05) compared with the plant species of the botanical groups of legumes and other forbs observed in the present study, was due to the fact that the stems of the grasses contain a higher percentage of crude fiber than the other botanical groups, as well as a higher stem/leaf ratio (  (Buxton, 1996;Ganskopp & Bohnert, 2001;González-Andrés & Ortiz, 1996). The lower NDF and ADF content of plants in the early stages of growth and its progressive increase as the plants grow has been reported by several investigators (Buxton & Redfearn, 1997;Tan et al., 2003).
The highest average content of the plant species of the botanical groups of legumes and other forbs in minerals, with statistically significant differences (P < .05) compared with the plant species of the botanical group of grasses, except for Mn, is due to the fact that the leaves have a higher mineral content than the stems (Myrvang et al., 2016;Tan et al., 1997) and that grasses have a higher stem/leaf ratio than other botanical groups (Buxton & Redfearn, 1997;Cherney, Mertens, & Moore, 1990 Note. Mean values followed by different letters (a and b), in the same row, differ significantly (P < .05). Mean values followed by different exponents ( 1, 2, 3, 4, 5, and 6 ), in the same column, differ significantly (P < .05).
T A B L E 1 1 Copper (Cu) content of plant species, at various stages of growth (mg kg −1 dry matter) Note. Mean values followed by different letters (a and b), in the same row, differ significantly (P < .05). Mean values followed by different exponent ( 1 and 2 ), in the same column, differ significantly (P < .05). Wivstad, 2014;Wyss & Kessler, 2002). Concerning the higher Mn content of the plant species of the botanical group of grasses compared with the plants of the other botanical groups, our results are in accordance with those of Hussain and Durrani (2008) and are in disagreement with those of Tufarelli et al. (2010a) and Pirhofer-Walzl et al. (2011).
The statistically significant differences (P < .05) observed in the mineral content of the plants studied during the various stages of growth are due to the fact that the leaves contain a higher percentage of minerals than the stems and that as the plants grow, leaf/stem ratio decreases (Arzani et al., 2004;Bovolenta, Spanghero, Dovier, Orlandi, & Clementel, 2008). Concerning the decrease in the mineral content observed in most of the plant species of the "Kostilata" grassland as the plants mature, the results of our study agree with those of Pirhofer-Walzl et al. (2011) and Schlegel et al. (2016). Different mineral contents between the different plant species, between the same plant species in different environments, and between the same plant species at different stages of growth have also been reported by other researchers (Elgersma & Søegaard, 2016;Lindström et al., 2014;Schlegel et al., 2016).
The mean NDF content of the plant species of the botanical group of grasses in the "Kostilata" grassland was lower than that of eight plant species of the botanical group of grasses in an experimental field in Italy, at an altitude of 1,500 m, whereas the average ADF content of the grasses in "Kostilata" grassland was in the intermediate range.
The mean NDF and ADF content of the plant species of the botanical group of other forbs of "Kostilata" grassland was higher than that of three plant species of the botanical group of other forbs in the above experimental field. The mean NDF content of P. alpinum L. in the "Kostilata" grassland was lower than that of P. alpinum L. in the experimental field of Italy, whereas the ADF contents of the two plant species were similar. The mean NDF and ADF content of T. repens L. in "Kostalata" grassland was higher than that of T. repens L. in the Italian experimental field (Bovolenta et al., 2008).
The mean NDF and ADF content of S. pennata L. in the "Kostalata" grassland was lower than that reported by Hussain and Durrani (2009). The mean NDF content of T. repens L. in the "Kostilata" grassland compared with the NDF content of T. repens L. at the flowering stage in a grassland in Turkey was higher, whereas its ADF content was lower (Kiraz, 2011). The mean NDF content of L. aphaca L. in the "Kostilata" grassland was higher than that of L. aphaca L. in grasslands of Southeastern Anatolia Region of Turkey (Basbag, Aydin, Çaçan, & Sayar, 2012). At a grassland in Italy at an altitude of 1,300 to 1,500 m, the values of NDF and ADF content of four legumes were slightly lower than those of the "Kostilata" grassland legumes. The values of NDF and ADF contents of the other forbs of corniculatus L. and A. millefolium L. showed lower NDF and ADF values than those of the corresponding plant species of "Kostilata" grassland (Tufarelli, Cazzato, Ficco, & Laudadio, 2010b). The NDF and ADF content of the other forbs of "Kostilata" grassland compared with the content of five other forbs in a grassland in Rieti, central Italy, at altitudes between 400 and 500 m, showed intermediate values (Tufarelli et al., 2010a). In May, the NDF and ADF contents of T.
repens L. and L. corniculatus L. of "Kostilata" grassland were higher than those of the corresponding plant species at Research Farm Foulumgaard in Denmark (Elgersma & Søegaard, 2016), and these differences can probably be attributed to different climatic and soil conditions.
The Ca and K contents of the grasses of "Kostilata" grassland were higher than those of lower altitude grasslands in northwestern Greece, near the area of our research, whereas the P and Mg contents were similar. The legumes of "Kostilata" grassland had a higher Ca and K content, whereas the P and Mg contents were lower. The other forbs of "Kostilata" grassland had higher Ca content, whereas the P, Mg, and K contents were similar (Roukos et al., 2011).
The content of all nine plant species of "Kostilata" grassland in Regarding Na, the grasses of "Kostilata" grassland had slightly higher content and the legumes and other forbs also had a higher content (Tufarelli et al., 2010a(Tufarelli et al., , 2010b. Concerning the average content of grasses in Mg, Ca, and K in relation to the average content of legumes and other forbs, our results are in agreement with those reported by Elliot (1986)  reported by Hussain and Durrani (2008).
Concerning variations in the content of plant species in Ca, the results of our study agree with those of Tan et al. (2003) with respect to A. gerardii V., S. pennata L., and G. lucidum L. As regards the P con- The average Ca, P, and Mg contents of L. aphaca L. in the "Kostilata" grassland were lower than those in a grassland in Turkey (Basbag et al., 2012).
The differences observed in both the content and the variation in the chemical composition between the plant species of the "Kostilata" grassland, compared with other plant species from other grasslands, can be mainly attributed to the different climatic conditions prevailing between the ecosystems (Flisch et al., 2009;Georgievskki et al., 1982;Öborn et al., 2008;Ramírez-Pérez et al., 2000).
The nine plant species we have studied, namely, the three most predominant of each botanical group, constitute the bulk of the grazing material of the grassland under investigation. Their nutritional value is good, as their NDF content does not exceed 600 g kg −1 DM during the study period. When the NDF content of forage exceeds 600 g kg −1 DM, there is a decrease in food consumption by animals (Mertens, 1994). Also, the mineral content of almost all plants meets the nutritional requirements of sheep, according to National Research Council (1985). Specifically, all plant species meet the nutritional requirements of the animals for K and Na. All plant species meet the needs of animals for Ca, except for P. alpinum L. Only other forbs and legumes meet the maintenance requirements in Mg, whereas only grasses and other forbs meet the requirements in P. As far as micronutrients are concerned, all plants meet the animal requirements for Fe and Mn, whereas no plants meet the requirements for Cu.
Among the plant species, those belonging to the botanical groups of legumes and other forbs have the best nutritional value. Therefore, proper grazing management should be applied to "Kostilata" grassland, so that the above botanical groups are preserved in the vegetation composition. According to Papanastasis et al. (2002), overgrazing of sheep in subalpine grasslands reduces plant diversity, especially in other forbs and legumes.

| CONCLUSION
The

ACKNOWLEDGEMENTS
This study was supported and cofinanced by the European Union (European Fund of Regional Development) and National resources in the framework of the Regional Operational Program "Thessaly -Central Greece -Epirus."

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.