Relationships between long-term fertilization management and forage nutritive value in grasslands

Relationships between long-term fertilization management and forage nutritive value in grasslands

Agriculture, Ecosystems and Environment 279 (2019) 139–148 Contents lists available at ScienceDirect Agriculture, Ecosystems and Environment journal...

2MB Sizes 0 Downloads 24 Views

Agriculture, Ecosystems and Environment 279 (2019) 139–148

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment journal homepage: www.elsevier.com/locate/agee

Relationships between long-term fertilization management and forage nutritive value in grasslands

T

Anna Dindováa, Josef Hakla, , Zuzana Hrevušováa, Pavel Nerušilb ⁎

a b

Department of Agroecology and Crop Production, Czech University of Life Sciences Prague, Kamýcká 129, 165 21 Praha 6, Suchdol, Czech Republic Research Institute of Crop Production Prague 6 – Ruzyně, Grassland Research Station Jevíčko, K.H. Borovského 461, 569 43 Jevíčko, Czech Republic

ARTICLE INFO

ABSTRACT

Keywords: Arrhenatherion elatioris Nitrogen Yield Digestibility NIRS

Effects of fertilization on grassland have been widely investigated in terms of botanical composition, forage yield and quality, and their agricultural and environmental consequences. However, there are no published studies of the contribution of fertilization to variability of forage quality involving systematic investigation of this effect. Therefore, our objectives were to: (i) evaluate the impact of long-term fertilization treatments on Arrhenatherion elatioris-type meadow vegetation, (ii) investigate effects of interaction among year, cut and fertilization on forage quality in association with forage yield and functional groups proportion with quantification of fertilization effect, and (iii) compare the predictive ability of visually estimated functional groups coverage (FGC) vs. weight proportion ratio of functional groups (FGW) for variability in forage quality in the first cut. Six mineral fertilization treatments (control, N0P40K100, and P40K100 in combination with 50, 100, 150 and 200 kg N ha−1) were evaluated over a three-year period (2014–2016) under a three-cut regime. Intensive NPK fertilization doubled forage yield and increased the grass proportion with a corresponding decrease of forbs and legumes. Grass proportion in harvested forage was positively correlated with fibre content in the first cut and with organic matter digestibility and net-energy for lactation in subsequent cuts. In all cuts there was a consistent correlation in ash and crude protein content with legumes and forbs. Fertilization had less influence on forage quality, compared with inter-year effects or seasonal variability. For each cut, fertilization contributed around 20% of the variability in forage quality, where half of its contribution was realized through increased yield and changes in proportion of functional groups. The explanation power of functional groups in the first cut was slightly higher for the FGC method than the FGW proportion method, which suggests both methods are at least as effective in relation to grassland nutritive value. In summary, if fertilization results in only small changes in grassland yield or botanical composition, its effect on forage quality is greatly reduced. Plant developmental stage or nutritive value of particular species are potential explanations for the rest of the variability.

1. Introduction In many parts of Europe livestock production is based on use of intensive grass leys, maize and grain-based supplementation, while semi-natural grasslands are often poorly utilized. Increasing the use of semi-natural, species-rich grasslands for grazing or mown forage could serve as a way to bridge biodiversity, conservation and livestock production (French, 2017). The natural variability of grassland forage quality can be reduced by appropriate management, in which increased cutting frequencies provide forage with better nutritive value (Čop et al., 2009). Grasslands used for intensive production are also usually fertilized with high inputs of nitrogen and phosphorus to increase forage production and quality (Bruinenberg et al., 2002). Adequate management of permanent grassland, using appropriate levels of ⁎

fertilizers and choosing optimal harvest timing, allows improved yield and quality of forage, which can reduce the need for expensive concentrate feeds for optimizing feeding rations and environmental impact can also be limited (Dale et al., 2013). In meadows, the effects of fertilization have been widely investigated, including several long-term experiments in which significant responses have been reported in forage yield and quality (Schellberg et al., 1999; Honsová et al., 2007; Hrevušová et al., 2015). Regarding forage quality, a range of studies investigated the association between nutrient supply, soil nutrient status and forage nutritive value or chemical composition (Gierus et al., 2005; Hejcman et al., 2010). However, reasons for the effects of fertilization must be carefully investigated. Van Soest et al. (1978) noted that factors such as water or fertilization affect forage quality via plant development. For botanically

Corresponding author at: Faculty of Agrobiology, Food and Natural Resources, Kamýcká 129, 165 00 Praha, Suchdol, Czech Republic. E-mail address: [email protected] (J. Hakl).

https://doi.org/10.1016/j.agee.2019.01.011 Received 24 August 2018; Received in revised form 23 January 2019; Accepted 29 January 2019 Available online 24 April 2019 0167-8809/ © 2019 Elsevier B.V. All rights reserved.

Agriculture, Ecosystems and Environment 279 (2019) 139–148

A. Dindová, et al.

Fig. 1. The monthly sums of precipitation and temperature means during the experimental years, meteorological station Lukavec, Czech Republic.

referred to either in terms of species or functional group dominance expressed in terms of percentage cover, usually recorded before first harvest (Hejcman et al., 2007; Hrevušová et al., 2015). Assessments species cover may not correspond fully with the actual species weight ratio as calculated on a dry matter basis. This can cause a lower explanation power of species coverage in terms of forage nutritive value. There is, however, a lack of studies comparing coverage measures vs. weight ratio on forage quality, although the weight proportions approach may be more useful than estimates of cover in explaining variability of nutritive traits. In summary, while many studies have investigated the effects of fertilization on grassland yield and botanical composition, few such studies have evaluated the impact of fertilization on grassland forage quality in conjunction with understanding the relationship with yield and botanical composition. Therefore the objectives of this study were (i) to evaluate the effects of various long-term fertilization treatments on proportion of functional groups, forage DMY and quality of forage of the meadow vegetation type Arrhenatherion elatioris, (ii) to investigate effects of interaction of external factors (year, cut, fertilization) on forage quality in association with forage DMY and functional groups, and (iii) to compare the usefulness of the method of visual estimation of functional group coverage versus the weight proportion method for understanding variability in forage quality in the first cut. Clarifying these relationships could greatly improve understanding of how fertilization affects forage quality of species-rich grassland.

similar swards, there is generally an inverse relationship between forage yield and quality. Increase of grassland dry matter yields (DMY) under higher applications of nutrients (especially N) has been frequently documented for fertilization using mineral fertilizers (Čámská and Skálová, 2012; Hrevušová et al., 2015) and organic fertilizers (Duffková and Libichová, 2013; Duffková et al., 2015) and limiting the N input can reduce both herbage DMY and crude protein (CP) content. However, increased grassland DMY through mineral fertilization can also result in increased crude fibre (CF) content and consequently decreased in vitro organic matter digestibility (OMD) (Dale et al., 2013), and a lower net-energy for lactation (NEL) (Čop et al., 2009). Forage CP can be diluted with increasing herbage accumulation or remain constant in harvested biomass under enhanced N rates (Duffková et al., 2015; Schellberg et al., 1999). Fertilizer application also affects the grassland botanical composition. For example, tall grasses in grassland communities become more dominant under increased levels of N fertilization (Hejcman et al., 2007; Honsová, et al., 2007; Čámská and Skálová, 2012). Čop et al. (2009) found that the effect of fertilization treatment on dominance of functional groups was greater than the effect of different cutting treatments. Changes in grassland botanical composition are also associated with changes in forage OMD and chemical composition (Bruinenberg et al., 2002). Andueza et al. (2010) described changes in the nutritive value of permanent grasslands during the first growth cycle, and found that botanical composition and plant functional types were important factors in explaining these differences. Legume forages generally have higher CP content than grasses (Ergon et al., 2017); therefore, in grass-legume swards mineral-N fertilization can reduce the CP content of harvested grassland forage as a result of its effect in decreasing the proportion of legume forage in the sward (Dale et al., 2013). Andueza et al. (2010) showed that an increasing proportion of forbs (excluding legumes) in grasslands resulted in a reduction in OMD. Their study also suggests that the proportion of grasses appears to be an important determinant of OMD during early vegetation stages, whereas the proportion of forbs appears to be more important for late vegetation stages. Subsequently, Andueza et al. (2016) established relationships between dominant species and forage quality parameters of permanent grasslands over the first growth cycle. In addition to basic inter-species differences in forage quality, species from semi-natural grasslands vary in their phenology, and the current developmental stage may have considerable effects on quality in grass species and also in forbs (Duru, 1997). In many studies, the grassland botanical composition has been

2. Material and methods 2.1. Study site The long-term fertilization experiment was established on a moderately dry Arrhenatherion grassland (49°34'6"N, 15°11'49"E; elevation 485 m a.s.l.) in the central part of the Czech Republic, near the village Senožaty. The soil is a Cambisol with a sandy-loamy texture, and the site is well-drained, with average depth of the groundwater at 0.65 m. The long-term mean annual temperature is 7.0 °C and mean annual precipitation is 641 mm. The monthly temperature means and precipitations totals during the growing season in years 2014–2016 are shown in Fig. 1 (Lukavec meteorological station). The highest annual precipitation occurred in 2014. The growth period from June to October was the warmest (14.9 °C) and driest (278 mm) in 2015. In 2016, there was a drought period in August and September.

140

Agriculture, Ecosystems and Environment 279 (2019) 139–148

A. Dindová, et al.

Table 1 Description of fertilization treatments with annual nutrient rates (kg ha−1) and soil chemical properties in upper 0–20 cm layer analyzed in 2012. Fertilizer treatments

Mineral N:P:K applied

pH (CaCl2)

K (mg kg−1)

P (mg kg−1)

Mg (mg kg−1)

Ca (mg kg−1)

Control PK 50NPK 100NPK 150NPK 200NPK

0:0:0 0:40:100 50:40:100 100:40:100 150:40:100 200:40:100

4.72 4.52 4.67 4.86 4.77 4.42

111 115 107 96 89 91

14 29 22 29 19 24

200 133 148 123 105 136

1661 1407 1796 1936 1862 1412

2.2. Experimental design and soil analysis

calculated from the fresh biomass yield and the dry matter content of the sub-sample from each particular plot. All dried samples (oven-dried at 60 °C to constant weight) were milled to pass through a 1-mm screen and were scanned at the Grassland Research Station Jevíčko with a Foss NIR System 6500 equipped with a spinning sample module, in reflectance range 1100–2500 nm, band width 2 nm, measured in small ring cups, with samples being scanned twice. The estimated qualitative traits (in g kg−1) were CP, fat, CF, ash, OMD (%), and NEL (MJ kg−1). Local calibration equations were developed for grasslands in the Czech Republic and their prediction accuracy is described by Míka et al. (2003).

In 1976, an experiment was established with six different treatments of mineral fertilization: unfertilized control (N0 P0 K0), PK (P40 K100) and combinations of P and K fertilization with increasing N input (100, 200, 300, 400 kg ha−1). In 1991 the N rates were reduced by half (50, 100, 150, 200 kg ha−1). The experiment was arranged in a completely randomized block design with four replicates (24 experimental plots, each plot was 3 x 4 m). The nitrogen was supplied once in the spring in the form of calcium ammonium nitrate (27.5% N). Phosphorus was applied in autumn as superphosphate (8.5% P) and potassium was supplied as potassium chloride (50% K). Acronyms of fertilization treatments with rates of applied nutrients are shown in Table 1. In 2012, soil samples were collected from the upper 20 cm soil layer. Soil exchangeable pH (CaCl2) was determined and plant available nutrients were extracted with a Mehlich III solution. The soil had acid reaction (pH 4.4–4.9) and the concentration of plant available P was generally below the optimum for productive grassland. The concentrations of K, Mg and Ca were not limiting for grass growth (Table 1).

2.5. Data analysis A three-way repeated measurement ANOVA was used (year, cut, fertilization) to evaluate the effects of fertilization treatments, cut, and year, on forage quality, DMY, and FGW. For better understanding of seasonal effect, two-way repeated measurement ANOVA (year, fertilization) was applied to detect the influence of fertilization treatments within each cut across the three years. Assumption for repeated measurement ANOVA were tested by Levene’s test of homogeneity of variance and Mauchly’s sphericity test. All factors were considered as fixed and significant differences between means were reported using the Tukey HSD test at α = 0.05. All these analyses were carried out using the STATISTICA program (StatSoft, 2012). Redundancy analysis (RDA) was used to perform analyses for effects across years and cuts as well as the variance partitioning procedure within each cut. This allows assessment of the proportion of variability of grassland forage quality that could be explained by explanatory variables, when the effect of others could be excluded as covariates. The option of centre and standardization by dependent variables was used. The statistical significance of the first and all of the other constrained canonical axes was determined by the Monte Carlo permutation test (499 permutations). All ordination analyses were performed in the CANOCO 4.5 program (ter Braak and Šmilauer, 2002). An ordination triplot was created in CanoDraw (Microcomputer Power, Ithaca, NY). The ordination triplot visualized the relationship between nutritive value traits (dependent variables) and year, cut order and fertilization (explanatory variables) where DMY, weight proportion of grasses, legumes and forbs were used as supplementary variables.

2.3. Functional groups and plant species composition In 2014–2016, cuts were taken in early June, mid-August and early October. In each plot, the sward was mown (using a mower with a cutter bar of 1.4 m working width) leaving a stubble height of approximately 5 cm. Weight percentage ratio of functional groups (FGW) was determined for each cut. A representative herbage sample (of approximately 0.5 kg) was collected from each plot. Each sample was divided into two subsamples: one was used for FGW assessment, and the other was kept for dry matter content and subsequent nutritive value measurements. The forage subsamples for determination of FGW assessment were hand-sorted into the three functional groups: grasses, legumes and nonleguminous dicotyledonous species (forbs). Senescent biomass was removed from samples. Samples were oven-dried at 60 °C to constant weight and FGW was calculated on a dry matter basis. The percentage cover of all vascular species was estimated visually before the first cut in early June of 2014, 2015 and 2016. In each plot, two 1 x 1 m relevés were recorded (eight relevés per treatment). Species nomenclature follows Kubát et al. (2002). Estimated coverage of the functional groups (FGC) was calculated from coverage of particular species within each group.

3. Results

2.4. Dry matter yield and forage quality

3.1. Species composition and seasonal changes in functional groups and forage yield

Fresh biomass yield was determined from the mown central area of 5.6 m2 within each plot. The DMY for each cut and season was

The number of species present prior to the first cut ranged from 50 in 2015 to 55 in 2014. The most dominant species in descending order

141

Agriculture, Ecosystems and Environment 279 (2019) 139–148

A. Dindová, et al.

Table 2 Mean functional groups cover (D, %), number of species with coverage ≥ 3% (N), Latin name abbreviations and cover of dominant species (DS, %) in each functional group before first cut. TrifRep – Trifolium repens, TrifPra – Trifolium pratense, LotuCor – Lotus corniculatus, LeonHis – Leontodon hispidus, PlanLan – Plantago lanceolata, AlchSp – Alchemilla sp., AnthSyl – Anthriscus sylvestris, TaraSp – Taraxacum sect. Ruderalia, RumeAce – Rumex acetosa, FestRub – Festuca rubra, PoaPra – Poa pratensis, ArrhEla – Arrhenatherum elatius, HolcLan – Holcus lanatus. Year

Treatment

Legumes

Forbs

Grasses

D

N

DS

D

N

DS

D

N

DS

2014

Control PK 50NPK 100NPK 150NPK 200NPK

31 31 21 9 7 2

3 4 3 2 1 0

TrifRep 10 TrifRep 12 TrifRep 7 TrifPra 4 TrifPra 3 TrifPra 1

38 33 35 30 25 26

3 4 2 5 2 2

LeonHis 11 PlanLan 6 AlchSp 7 AlchSp 4 TaraSp 4 AnthSyl 7

31 36 44 61 68 72

5 5 5 7 5 8

FestRub 5 PoaPra 9 ArrhEla 8 ArrhEla 17 ArrhEla 21 ArrhEla 21

2015

Control PK 50NPK 100NPK 150NPK 200NPK

23 20 11 5 4 1

3 3 1 0 0 0

LotuCor 8 TrifPra 8 TrifPra 4 TrifPra 2,5 TrifPra 2 TrifPra 0,5

40 31 27 20 18 22

4 2 3 1 0 1

LeonHis 12 LeonHis 5 RumeAce 4 AlchSp 3 RumeAce 2,5 AnthSyl 6

37 49 61 75 78 77

6 6 8 6 5 6

HolcLan 7 HolcLan 10 HolcLan 16 ArrhEla 18 ArrhEla 25 ArrhEla 28

2016

Control PK 50NPK 100NPK 150NPK 200NPK

24 21 12 4 3 1

3 4 1 0 0 0

LotuCor 9 TrifPra 8 TrifPra 5 TrifPra 1 TrifPra 1 LotuCor 0,5

34 25 19 16 12 13

3 2 1 1 0 1

LeonHis 12 LeonHis 5 AlchSp 3 AlchSp 3 TaraSp 2 AnthSyl 5

42 54 69 80 85 86

6 7 8 7 6 6

FestRub 7 PoaPra 10 HolcLan 18 ArrhEla 19 ArrhEla 29 ArrhEla 34

were Arrhenatherum elatius, Holcus lanatus, Poa pratensis, Trisetum flavescens, Alopecurus pratensis, Trifolium pratense, Leontodon hispidus and Lotus corniculatus. Mean cover values of the grasses, forbs and legumes at particular treatments in each year are shown in Table 2, where only slight changes in dominant species or species number with coverage ≥ 3% were observed within each fertilization treatments over the threeyear period. Weight proportions of functional groups were influenced by year, cut, and fertilization, as shown in Table 3. The lowest grass ratio was in 2014 (653 g kg−1) whereas significantly higher values of around 800 g kg-1 were observed in 2015 and 2016. Inversely, legumes and forbs achieved the highest proportions in 2014 with values 91 g kg−1 a 256 g kg−1, respectively. In 2015 and 2016, values of legumes and forbs decreased to an average of 55 g kg-1 and 142 g kg-1, respectively. The first cut provided a significantly higher grass proportion of 826 g kg-1 compared to other cuts. Inversely, the proportions of legumes and forbs increased in the second and third cuts.

Year and cut significantly affected grassland DMY (Table 3). The annual DMY was higher in 2014, the year with the highest annual precipitation, and lowest in 2016, the year with a drought period in August and September. There was a significant effect of cut on grassland DMY, which decreased significantly from the first to the third cut in all years. Effect of fertilization on DMY was highly significant (P = 0.001); the unfertilized plots provided annual DMY of only 4 t ha−1 compared with 8-10 t ha−1 from the N-fertilized plots. 3.2. Variability of forage nutritive value in years and cuts Differences in forage quality between years and cuts are summarized in Table 4. There were significant variations between years for CP, ash, NEL and OMD; the year 2015 exhibited the highest values NEL and OMD, together with the lowest value of CP and ash. The highest CP content was observed in 2014. In 2016, all nutritive values varied within the range of previous years. Values for CP, fat and NEL were

Table 3 Effect of year and cut on annual dry matter yield (DMY, t ha−1) and proportion of functional group as legumes (L), forbs (F), and grasses (G) in g kg−1. Significance of fertilization effect is reported in last column. Year

DMY L F G

Cut

Fertilization

2014

2015

2016

P

1

2

3

P

8.51 91a 256a 653b

7.98 52b 141b 807a

7.38 57b 143b 800a

0.242 0.006 < 0.001 < 0.001

5.19a 43b 131b 826a

1.97b 93a 210a 697b

0.80c 64ab 198a 738b

< 0.001 0.001 0.005 0.001

P 0.001 < 0.001 < 0.001 < 0.001

Three-way ANOVA, P – value represents significance of each factor on all tested variables, different letters document statistical differences within each factor for Tukey HSD, α = 0.05.

142

Agriculture, Ecosystems and Environment 279 (2019) 139–148

A. Dindová, et al.

Table 4 Effect of year and cut on crude protein (CP), fat, crude fibre (CF), ash (g kg−1), organic matter digestibility (OMD, %) and net-energy for lactation (NEL, MJ kg−1). Significance of fertilization effect is reported in last column. Year

CP Fat CF Ash NEL OMD

Cut

Fertilization

2014

2015

2016

P

1

2

3

P

P

145a 28 255 101a 5.12b 61.9c

123b 26 259 84c 5.55a 65.6a

136a 28 258 94b 5.20b 63.3b

< 0.001 0.402 0.740 < 0.001 < 0.001 < 0.001

116c 19c 288a 87b 4.83c 62.5b

128b 28b 254b 90b 5.44b 62.6b

160a 35a 230c 103a 5.59a 65.6a

< < < < < <

0.001 0.001 0.001 0.001 0.001 0.001

< 0.001 0.180 < 0.001 < 0.001 0.483 0.668

Three-way ANOVA, P – value represents significance of each factor on all tested variables, different letters document statistical differences within years or cuts for Tukey HSD, α = 0.05.

significantly increased whereas CF significantly decreased from first to third cut. Values for ash and OMD were higher in the third cut than in cuts 1 and 2. Fertilization significantly influenced CP, CF and ash in harvested forage.

coverages of legumes and forbs were higher in comparison with their weight ratio. Effect of fertilization on forage nutritive value, separately for each cut, is shown in Table 6. Consistent significant increase of CF under increasing rates of N is clearly visible in each cut, together with a slight tendency for ash reduction. In spite of increase in CF, NEL was significantly reduced in comparison with the control only in the first cut, and OMD was not significantly influenced by fertilization treatments. Higher CP content corresponds with higher rates of N in the first and third cut. In the second and third cuts, there was a positive effect of NPK fertilization on fat content when compared with the control. Ash content was reduced with higher N rates and this effect was the most visible in the second cut.

3.3. Effect of fertilization on proportion of functional groups and forage nutritive value within cuts Variation of FGW and DMY under fertilization treatments within each cut is documented in Table 5. There was a positive significant response of spring N fertilization on DMY in the first and second cut. Nitrogen fertilization was also associated with an increased proportion of grass and a decrease in legumes and forbs, and this effect was consistent in all three cuts. Percentage FGC correspond with changes in FGW; however, estimated grass coverage was lower whilst estimated

Table 5 Effect of cut and fertilization treatments on dry matter yield (DMY, t ha−1) and average values for the proportion of legumes (L), forbs (F), and grasses (G) in g kg−1 averaged across the years 2014–2016. Values in bracket represent estimated percentage cover of functional groups. Cut

Treatment

P

control

PK

50NPK

100NPK

150NPK

200NPK

1

DMY L F G

2.71c 88a (26) 317a (38) 595c (36)

3.76b 97a (25) 164b (29) 739bc (46)

6.11a 30b (14) 76b (28) 894ab (58)

6.21a 19b (6) 95b (22) 886ab (72)

6.33a 20b (5) 73b (18) 908a (77)

6.05a 5b (1) 62b (20) 934a (79)

< < < <

0.001 0.001 0.001 0.001

2

DMY L F G

1.18c 209a 536a 255c

1.68bc 186ab 263b 551b

1.85bc 124b 204bc 672b

2.11ab 34c 107cd 589a

2.68a 1c 89cd 610a

2.30ab 6c 61d 934a

< < < <

0.001 0.001 0.001 0.001

3

DMY L F G DMY

0.39 138a 386a 476c 4.28c

0.76 124a 263b 613b 6.20b

0.85 80ab 167bc 753a 8.81a

0.87 21bc 138c 841a 9.19a

0.90 14c 126c 861a 9.91a

1.3 5c 112c 884a 9.65a

< < < <

0.256 0.001 0.001 0.001 0.001

Annual

Two-way ANOVA, P – value represents significance of each factor on all tested variables, different letters document statistical differences within each row for Tukey HSD, α = 0.05.

143

Agriculture, Ecosystems and Environment 279 (2019) 139–148

A. Dindová, et al.

Table 6 Effect of cut and fertilization treatments on crude protein (CP), fat, crude fibre (CF), ash (all in g kg−1), organic matter digestibility (OMD, %) and net-energy of lactation (NEL, MJ kg−1) averaged across the years 2014–2016. Cut

Nutritive value

Treatment control

P PK

50NPK

100NPK

150NPK

200NPK

1

CP Fat CF Ash NEL OMD

115b 17 260c 92 5.12a 62.9

112b 20 277bc 93 4.87b 62.4

100b 19 294ab 86 4.77b 62.9

109b 19 297ab 86 4.72b 63.5

113b 17 309a 82 4.66b 61.9

146a 20 294ab 84 4.82b 61.5

< 0.001 0.598 < 0.001 0.054 < 0.001 0.614

2

CP Fat CF Ash NEL OMD

130 24c 230c 96ab 5.52 62.9

128 29abc 248b 99a 5.34 62.0

131 32a 247b 96ab 5.46 63.5

123 30ab 263ab 86ab 5.37 62.8

119 26bc 270a 81b 5.45 62.8

139 26bc 268a 82b 5.52 61.6

0.151 0.001 < 0.001 0.003 0.736 0.345

3

CP Fat CF Ash NEL OMD

156b 31b 216b 107 5.54 65.8

156b 35ab 230ab 107 5.47 65.3

154b 36ab 236a 103 5.48 65.6

152b 37a 236a 99 5.62 65.6

162b 37a 234ab 99 5.66 65.9

178a 37a 231ab 101 5.76 65.7

< 0.001 0.004 0.016 0.148 0.341 0.999

Two-way ANOVA, P – value represents significance of each factor on all tested variables, different letters document statistical differences within each row for Tukey HSD, α = 0.05.

3.4. Relationships among external factors and forage nutritive value in association with yield and functional groups ratio

canonical axes). These relationships among nutritive value, year and cut order are illustrated in the ordination triplot of RDA (Fig. 2). The most important first (horizontal) canonical axis represents the cut effect (with shift from right to left) explaining 43.3% of variability. The effect of cut was given by inverse relationship between high CF and reduction of other nutritive traits. The second (vertical) axis represents year, where it explained about 14% of variability of qualitative traits, with the highest difference between 2015 (top) and 2014 (bottom) and 2016 is located in the centre of the Figure. This corresponds with ANOVA results with the highest OMD and NEL in 2015 (Table 4). Supplementary variables show a trend for increase in CF and DMY with a higher grass ratio. Higher grass ratio was also positively related to improved OMD and NEL on the second canonical axis. Legumes and forbs were positively correlated with CP and ash but they showed a tendency for reduced DMY (first axis) and lower OMD and NEL (second axis). Fertilization management points are located in the centre of the figure;

The contribution of tested factors to total variability of forage nutritive value was investigated and the results of RDA are summarized in Table 7. First, the multivariate analysis investigated the contribution of year, cut, and fertilization to nutritive traits. In the analyses, the tested factors explained 64.9% of the variability of nutritive traits (all Table 7 Results of redundancy analyses investigating effect of explanatory variables on forage quality in each cut (FQ = forage quality; DMY = dry matter yield; Y = year; C = cut; F = fertilization; FGW = functional groups weight; FGC = functional groups cover; traits included as functional groups and forage quality are shown in Tables 3 and 4, respectively). Tested variables

Explanatory variables

covariate

% ax.1 (all)†

F1 (all)‡

P1 (all)§

FQ

Y, C, F (Fig. 2) F



F

Y, DMY, FGW

F

Y, DMY, FGC

FGC

Y

FGC

Y, DMY

F

Y

F

Y, DMY, FGW

F

Y

F

Y, DMY, FGW

43.3 (64.9) 14.3 (22.2) 6.0 (9.7) 4.0 (7.5) 11.3 (11.7) 3.8 (4.2) 12.4 (21.1) 5.3 (9.9) 9.0 (15.5) 5.9 (8.1)

157.5 (42.2) 13.9 (4.9) 6.1 (2.1) 4.2 (1.7) 11.1 (5.7) 3.9 (2.2) 16.9 (7.1) 8.3 (3.5) 10.8 (4.2) 7.8 (2.2)

0.002 (0.002) 0.002 (0.002) 0.080 (0.020) 0.272 (0.088) 0.002 (0.002) 0.040 (0.046) 0.002 (0.002) 0.008 (0.002) 0.002 (0.002) 0.010 (0.008)

FQ

FQ

FQ

1st cut

2nd cut

3rd cut

Y



% ax.1 (all) – variability of grassland forage quality explained by canonical axis 1 or by all axes in brackets. ‡ F 1 (all) – F statistics for the test of axis 1 or all axes in brackets. § P 1 (all) – corresponding probability value obtained by the Monte Carlo permutation test (499 permutations) for the test of axis 1 or all axes in brackets.

Fig. 2. Ordination diagram showing result of RDA testing the influence of fertilization treatment, cut and year on grassland forage quality. Functional group weight proportion and dry matter yield were used as supplementary variables. 144

Agriculture, Ecosystems and Environment 279 (2019) 139–148

A. Dindová, et al.

Fig. 3. Explanation power of fertilization (F, %) to variability of forage nutritive value followed by percentage contribution of dry matter yield (DMY), proportion of functional groups weights (FGW) and their overlap within fertilization effect in each cut.

145

Agriculture, Ecosystems and Environment 279 (2019) 139–148

A. Dindová, et al.

therefore the fertilization effect was marginal with regard to cut and year. For this reason, variation partitioning was analysed separately within each cut, in which the effect of year was excluded as a covariate.

4.2. Influence of long-term fertilization on forage nutritive value Across years and cuts, effect of fertilization was significant for CP, ash and CF content. Within each cut, only CF was consistently increased by N fertilization, which was consistent with results of Čop et al. (2009). Ash content increased from first to third cut in line with Mohammed et al. (2016). Fat content showed a tendency to be higher with N fertilization. Forage CP concentration was significantly higher in the 200NPK treatment compared with other treatments in the first and third cuts. This finding is in line with Schellberg et al. (1999), that fertilizer application has direct effects on the herbage mineral concentration, including N content in grassland species. Lemaire et al. (2008) described a N-dilution curve with increasing biomass yield in association with leaf area index and N uptake changes. In line with this, a stagnation of N concentration and a significant increase of biomass yield were observed with increasing slurry fertilization on moderately dry Arrhenatherion grassland by Duffková et al. (2015). This stagnation of N concentration was visible up to the rate 150 kg N in the present experiment, and then the CP concentration significantly increased simultaneously with yield stagnation between 150NPK and 200NPK. This provides evidence that tall-grass dominated grasslands could accumulate an excess of N from fertilization and is in line with results of Brum et al. (2009). It must be taken into account that the non-protein N fraction (NPN) in forage may account for a considerable part of the increase total CP content under N fertilization. This increase of NPN can be partly reduced by sulphur application, where NPN content achieved values of 15% without S application and decreased to 9% with the combined N and S application (Gierus et al., 2005). Higher NPN under intensive N fertilization can also help explain the disputable effectiveness of higher forage CP content in terms of increasing milk production, and the recommendation, based on this, that grassland N fertilization should be optimized according to expected yield (Huhtanen and Broderick, 2016).

3.5. Variation partitioning of fertilization effect on forage quality Variation partitioning by RDA reveals that fertilization significantly explained about 20% of forage qualitative traits variability in each cut (Table 7). About 50% of the fertilization contribution could be explained by changes in DMY and variation in FGW; Fig. 3 shows these contributions as well as an overlap of their respective influences. The contribution of DMY and FGW fluctuated considerably among cuts, and the importance of DMY was reduced from 11.6% in the first cut to 2.8% in the third cut. On the other hand, the contribution of FGW was relatively sTable (5.9–10.4%) but its overlap with DMY was also found to be decreasing from first to third cut. Fertilization effect was reduced by half after excluding the DMY and FGW effect but it remained significant and contributed up to 9% of explained variability. A comparison of the contributions of FGC and FGW on forage quality was realised in the first cut (Table 5). Cover of functional groups reached higher explanation power (11.7%, see Table 7) than FGW (8.6%, see Fig. 3) and its effect on forage quality was still significant after excluding the DMY effect. Using FGC as a covariate reduced the significance of the fertilization effect. 4. Discussion 4.1. Variability in grassland species composition, functional groups, and forage yield The 50 to 55 vascular plant species recorded in the period 2014–2016 represent the common species of Arrhenatherion elatioris meadow vegetation (Chytrý, 2007) and this was similar to other studies conducted in Arrhenatherum meadows (Čop et al., 2009; Čámská and Skálová, 2012). Different long-term fertilization management resulted in significant changes in grassland species composition where Arrhenatherum elatius was the dominant grass species in N fertilized plots. Results also show a decrease in species number with increasing N fertilization for forbs and especially legumes. This is consistent with results from Beltman et al. (2007), where species richness correlated negatively with above-ground biomass. However, negative effects on species richness in Arrhenatherum meadows may not be visible in waterlimited conditions (Duffková et al., 2015). The effect of fertilization on functional groups proportion was visible across cuts and years. Increased proportion of grass under increasing N fertilization and consequent reductions in legume and forb proportions corresponds with the previously well-documented pattern obtained in a range of studies in Arrhenatherum meadows (Čámská and Skálová, 2012; Duffková and Libichová, 2013). Average proportion of functional groups is in line with Čop et al. (2009) for a three-cut regime. Decrease of grasses and increase of legumes and forbs over the season was consistent with Michaud et al. (2011). It can be explained by faster growth of grasses in response to N fertilization in the spring period (Brum et al., 2009). Although N application took place in early spring, differences in functional groups among fertilization treatments were clearly visible across cuts. The DMY values obtained in this study correspond with results showing that DMY of Arrhenatherum meadows can approximately double under intensive fertilization management (Čop et al., 2009; Duffková et al., 2015). Similarly, cessation of fertilization has been shown to lead to a decrease in annual DMY from 9 to 5 t ha−1 over a four-year period (Hofmann and Isselstein, 2005). Percentage proportion of the first, second, and third cuts of annual DMY correspond with values reported by Hrevušová et al. (2015).

4.3. Relationships among external factors and nutritive value traits The changes in grassland composition or nutritive traits as described above clearly document a typical Arrhenatherion grassland response to fertilization management. Based on this background, the seasonal effects, year and fertilization explained about 65% of the nutritive value variability. The most important trend was an increase of CF with a simultaneous decrease of all other nutritional traits (first canonical axis, 43% of explained variability). This effect was directly related to cut order over season from first cut (left) to third cut (right side of Fig. 2). These differences correspond with the seasonal development of forage quality described by Schellberg et al. (1999); Brum et al. (2009) or Mohammed et al. (2016). Results in the study of Schellberg et al. (1999) showed higher CP in the second cut in comparison with the first cut, which was in line with results of the present study. This may be explained by a higher proportion of N-rich leaves in the harvested biomass of lower-yield summer and autumn cuts. Seasonal pattern in association with DMY and functional groups can be considered as the factor primarily driving 2 3 of the variability of grassland forage quality. The impact of fertilization on forage quality was marginal, compared with cut or year effect; similarly the small effect of drought on forage quality was in contrast to the seasonal effect reported in the study of Küchenmeister et al. (2014). The vertical axis clearly separated the year 2014 with the highest values of CP and ash (top) from the year 2015 with the highest values of NEL and OMD. The explanation for these differences could not be attributed simply to different weather conditions. Although higher temperature increases lignification and thus reduces OMD (Van Soest et al., 1978) the differences reported here do not correspond with higher temperatures in the July-August period of 2015. Further, the OMD difference was not associated even with higher DMY because this value was not significantly different among years in spite drought periods in 2015 and 2016. The absence of 146

Agriculture, Ecosystems and Environment 279 (2019) 139–148

A. Dindová, et al.

significant differences in DMY was also in line with similar CF content between years. Based on Fig. 2, it seems that significant changes in forage quality could be associated with the proportions of different functional groups. The positive correlation between grass proportion and fibre content was consistent with results of Küchenmeister et al. (2014) where grass species had higher concentration of fibre and lower concentration of NEL in mixtures than in pure stands (Ergon et al., 2017). In the present study, higher proportions of legumes and forbs increased CP and ash content, whereas a higher proportions of grass was positively correlated with CF in the first cut and with OMD and NEL in the subsequent cuts. This is in line with the negative effect of forbs on forage OMD reported by Andueza et al. (2010). These results support the idea that increasing the dominance of grass in Arrhenatherion grassland can improve some components of forage quality, especially OMD and NEL in the second and third cut. It is also in line with Michaud et al. (2011) in terms of the positive impact of the dominance of productive grass species on forage quality in grassland with contrasting functional types. According to Küchenmeister et al. (2014), grasses increased water soluble carbohydrates (WSC) content, which corresponds with their significantly higher WSC content compared with legumes (Ergon et al., 2017). On the other hand, higher CP and ash was observed in 2014 suggesting potential of forbs and legumes for better CP and mineral supplementation. Bruinenberg et al. (2002) noted that dicotyledonous species may affect OMD in either a positive or negative way, depending on character of neutral detergent fibre or plant anatomy. Legumes exhibited the highest CP content of all functional groups in grassland (Schellberg et al., 1999). Reduced CP content with decreasing legume proportion has been reported by Dale et al. (2013). According to Hofmann and Isselstein (2005), the higher dominance of legumes supported an increase in feeding value in terms of the CP content in the forage, but potential benefit of this CP for enhanced milk performance must be questioned (Huhtanen and Broderick, 2016). Dale et al. (2013) also observed a positive impact of cow and horse manure application on forage OMD in association with higher yield and legume proportion.

covariate. Fertilization effect became non-significant after excluding the effect of year, DMY and FGC as a covariate. Higher contribution of FGC could be associated with higher estimated coverage of legumes and forbs rather than by their weight proportions in the biomass. The use of FGC with short and tall forbs and grasses did not substantially improve the explanation power in comparison with FGC alone (data not shown). In spite of some differences in proportion of functional groups, it seems that simply visually estimated coverage of the main functional groups in intensive grasslands could be equal to more time consuming FGW proportion method as a tool for explanation of differences in forage quality, especially in cases where there are similar DMY among fertilization treatments. 4.5. Contribution of yield and stand composition to explanation of fertilization effect Results of the present study show that changes in DMY can be considered as the most important factor driving variability of forage quality in the first cut where the highest differences among fertilization treatments occurred. This effect is in line with the well-known positive correlation between DMY and fibre content with consequently reduced OMD (Elgersma and Søegaard, 2018). There was a high overlap between DMY and FGW in the first cut where the FGW effect becomes non-significant after excluding the effect of DMY as a covariate. It is also supported by results of Michaud et al. (2011) where a slower decrease in OMD between grassland with contrasting functional types was explained by lower amounts of forage production. Contribution of stand composition increased if DMY differences were reduced among fertilization treatments in the second and third cuts. Therefore, FGW can be considered as the second most important factor explaining variability of forage quality with regard to grassland fertilization, and this is in line with the importance of functional groups for forage quality as described in range of studies (Andueza et al., 2010; Küchenmeister et al., 2014). The significance of fertilization after using DMY and FGW as covariates suggest that changes of grassland forage quality could not be attributed simply to the functional groups-yield relationship. Schellberg et al. (1999) observed some fluctuation in NEL and CP within functional groups across various fertilization managements, in line with particular species composition. Bruinenberg et al. (2002) reported high variability in forage quality among grassland species where competitive species are associated with high yield and forage quality value in comparison with species that are less demanding for nutrients and resources (Andueza et al., 2016). It can be concluded that the rest of the variability was probably connected with other factors such as the developmental stages of species within the same yield or nutritive value of particular species.

4.4. Functional group coverage vs. weight ratio response in forage quality Estimation of grassland species/functional groups coverage has been widely used as a traditional method for botanical composition assessment. In the present study, comparison in the first cut clearly showed that grass coverage was underestimated in comparison with true weight proportion, whereas the opposite was observed for dicotyledonous species. This trend was consistent across fertilization treatments and years. It is also in line with the accepted difficulty of estimating stand composition by visual observation alone, which was reported by Parsons et al. (2006) even for binary lucerne-grass mixtures. Due to inaccuracy of visual estimation, and the excessive time required for hand separation of binary mixtures, digital image analysis has been proposed by McRoberts et al. (2016). Visual estimation of species coverage has been used as the most common standard for assessment of grassland botanical composition where the focus of the research is on diversity (Čámská and Skálová, 2012; Hrevušová et al., 2015; Duffková et al., 2015) whereas the weight proportion of functional groups is generally preferred for studies where the emphasis is on forage nutritive value (Čop et al., 2009; Brum et al., 2009; Andueza et al., 2010; Küchenmeister et al., 2014). The relative weight proportions of particular species within herbage from cut swards have been investigated in relatively few studies (for instance, Andueza et al. (2016)). The importance of this topic was demonstrated in the study of Parsons et al. (2006) which showed there was greater accuracy of fibre prediction in a lucerne-grass mixture when determined by grass weight proportion than by estimation of grass fraction, although magnitudes of the errors were still at an acceptable level for both variables. In our study, explanation power of FGC was higher (4.2%, Table 7) than FGW (0.9%, calculated as 8.6 – 7.7%, Fig. 3) considering year and DMY as

5. Conclusion This study showed that the forage quality of Arrhenatherion grassland was primarily driven by seasonal patterns and inter-year differences. The changes in forage quality were strongly associated with differences in DMY and also functional groups. When the levels of DMY were higher this resulted in higher CF contents in the harvested herbage. The effect of proportion of grasses on overall forage quality varied over the duration of the growing season: where grass proportion increased then CF increased in the first cut, under increasing forage yield, but in the second and third cuts the increased presence of grasses improved forage OMD and NEL. Legumes and forbs consistently increased ash and CP content. Nutrient supply from fertilization accounted for around 20% of the variability of forage quality within a cut. Fertilization not only resulted in increased DMY but also led to changes in functional groups; together these represent about 50% of the variability explained by fertilization. The explanation power of functional groups to variability of forage quality in the first cut was slightly higher for the estimated FGC than for FGW method; this suggests that these 147

Agriculture, Ecosystems and Environment 279 (2019) 139–148

A. Dindová, et al.

two methods are at least equivalent in the prediction of grassland nutritional value.

harvest, and relationships between productivity and components of feed quality. Grass Forage Sci. 73 (1), 78–93. Ergon, Å., Kirwan, L., Fystro, G., Bleken, M.A., Collins, R.P., Rognli, O.A., 2017. Species interactions in a grassland mixture under low nitrogen fertilization and two cutting frequencies. II. Nutritional quality. Grass Forage Sci. 72 (2), 333–342. French, K.E., 2017. Species composition determines forage quality and medicinal value of high diversity grasslands in lowland England. Agric., Ecosyst. Environ., Appl. Soil Ecol. 241, 193–204. Gierus, M., Jahns, U., Wulfes, R., Wiermann, C., Taube, F., 2005. Forage quality and yield increments of intensive managed grassland in response to combined sulphur-nitrogen fertilization. Acta Agric. Scand. B-S. P. 55 (4), 264–274. Hejcman, M., Klaudisová, M., Schellberg, J., Honsová, D., 2007. The Rengen Grassland Experiment: plant species composition after 64 years of fertilizer application. Agric., Ecosyst. Environ., Appl. Soil Ecol. 122 (2), 259–266. Hejcman, M., Szaková, J., Schellberg, J., Tlustoš, P., 2010. The Rengen Grassland Experiment: relationship between soil and biomass chemical properties, amount of elements applied, and their uptake. Plant Soil 333 (1-2), 163–179. Hofmann, M., Isselstein, J., 2005. Species enrichment in an agriculturally improved grassland and its effects on botanical composition yield and forage quality. Grass Forage Sci. 60 (2), 136–145. Honsová, D., Hejcman, M., Klaudisová, M., Pavlů, V., Kocourková, D., Hakl, J., 2007. Species composition of an alluvial meadow after 40 years of applying nitrogen, phospohorus and potassium fertilizer. Preslia 79 (3), 245–258. Hrevušová, Z., Hejcman, M., Hakl, J., Mrkvička, J., 2015. Soil chemical properties, plant species composition, herbage quality, production and nutrient uptake of and alluvial meadow after 45 years of N, P, and K application. Grass Forage Sci. 70 (2), 205–218. Huhtanen, P., Broderick, G., 2016. Improving utilization of forage protein in ruminant production by crop and feed management. The Multiple Roles of Grassland in the European Bioeconomy. Proceeding of the 26th General Meeting of the European Grassland Federation 340–349. Kubát, K., Hrouda, L., Chrtek, J.jun, Kaplan, Z., Kirschner, J., Štěpánek, J., 2002. Klíč ke květeně České republiky. Academia, Praha. Küchenmeister, F., Küchenmeister, K., Kayser, M., Wrage-Mönnig, N., Isselstein, J., 2014. Effects of drought stress and sward botanical composition on the nutritive value of grassland herbage. Int. J. Agric. Biol. 16 (4), 715–722. Lemaire, G., van Oosterom, E., Jeuffroy, M.H., Gastal, F., Massignam, A., 2008. Crop species present different qualitative types of response to N deficiency during their vegetative growth. Field Crops Res. 105 (3), 253–265. McRoberts, K.C., Benson, B.M., Mudrak, E.L., Parsons, D., Cherney, D.J.R., 2016. Application of local binary patterns in digital images to estimate botanical composition in mixed alfalfa-grass fields. Comput. Electron. Agric. 123, 95–103. Michaud, A., Andueza, D., Picard, F., Plantureux, S., Baumont, R., 2011. Seasonal dynamics of biomass production and herbage quality of three grasslands with contrasting functional compositions. Grass Forage Sci. 67 (1), 64–76. Míka, V., Pozdíšek, J., Tillmann, P., Nerušil, P., Buchgraber, K., Gruber, L., 2003. Development of NIR calibration valid for two different grass sample collections. Czech J. Anim. Sci. 48 (10), 419–424. Mohammed, G., Trolard, F., Bourrié, G., Gillon, M., Tronc, D., Charron, F., 2016. A longterm data sequence (1960-2013) to analyse the sustainability of hay quality in irrigated permanent grasslands under climate change. Am. J. Agric. For. 4 (6), 140–151. Parsons, D., Cherney, J.H., Gauch, H.G., 2006. Estimation of preharvest fiber content of mixed alfalfa-grass stands in New York. Agron. J. 98, 1081–1089. Schellberg, J., Möseler, B.M., Kühbauch, W., Rademacher, I.F., 1999. Long-term effects of fertilizer on soil nutrient concentration, yield, forage quality and floristic composition of a hay meadow in the Eifel mountains, Germany. Grass Forage Sci. 54 (3), 195–207. StatSoft, Inc., 2012. Statistica for Windows. StatSoft, Tulsa, USA. Ter Braak, C.J.F., Šmilauer, P., 2002. CANOCO Reference Manual and CanoDraw for Windows User’s Guide: Software for Canonical Community Ordination (version 4.5). Microcomputer Power, Ithaca, USA. Van Soest, P.J., Mertens, D.R., Deinum, B., 1978. Preharvest Factors Influencing Quality of Conserved Forage. J. Anim. Sci. 47 (3), 712–720.

Funding This research was supported by “S” grant of MŠMT ČR and project MZE-RO0418 of Ministry of Agriculture of the Czech Republic. The completion of the paper was supported by project SV16-39-21240 of Grant Agency of the Faculty of Agrobiology, Food and Natural Resources, Czech University of Life Sciences Prague. Acknowledgements We sincerely thank all colleagues who helped with the laboratory work. References Andueza, D., Cruz, P., Farruggia, A., Baumont, R., Picard, F., Michalet-Doreau, B., 2010. Nutritive value of two meadows and relationships with some vegetation traits. Grass Forage Sci. 65 (3), 325–334. Andueza, D., Rodrigues, A.M., Picard, F., Rossignol, N., Baumont, R., Cecato, U., Farruggia, A., 2016. Relationships between botanical composition, yield and forage quality of permanent grasslands over the first growth cycle. Grass Forage Sci. 71 (3), 366–378. Beltman, B., Willems, J.H., Güsewell, S., 2007. Flood events overrule fertilizer effects on biomass production and species richness in riverine grasslands. J. Veg. Sci. 18 (5), 625–634. Bruinenberg, M.H., Valk, H., Korevaar, H., Struik, P.C., 2002. Factors affecting digestibility of temperate forages from seminatural grasslands: a review. Grass Forage Sci. 57 (3), 292–301. Brum, O.B., López, S., García, R., Andrés, S., Calleja, A., 2009. Influence of harvest season, cutting frequency and nitrogen fertilization of mountain meadows on yield, floristic composition and protein content of herbage. Rev. Bras. Zootecn. 38 (4), 596–604. Čámská, K., Skálová, H., 2012. Effect of low-dose N application and early mowing on plant species composition of mesophilous meadow grassland (Arrhenatherion) in Central Europe. Grass Forage Sci. 67 (3), 403–410. Chytrý, M., 2007. Vegetation of Czech Republic 1. Grassland and Heathland Vegetation. Academia, Praha 2007. Čop, J., Vidrih, M., Hacin, J., 2009. Influence of cutting regime and fertilizer application on the botanical composition, yield and nutritive value of herbage of wet grasslands in Central Europe. Grass Forage Sci. 64 (4), 454–465. Dale, L.M., Thewis, A., Rotar, I., Boudry, C., Pacurar, F.S., Lecler, B., Agneessens, R., Dardenne, P., Baeten, V., 2013. Fertilization effects on chemical composition and in vitro organic matter digestibility of semi-natural meadows as predicted by NIR spectrometry. Not. Bot. Horti Agrobot. Cluj. 41 (1), 58–64. Duffková, R., Libichová, H., 2013. Effects of cattle slurry application on plant species composition of moderately moist Arrhenatherion grassland. Plant Soil Environ. 59 (11), 485–491. Duffková, R., Hejcman, M., Libichová, H., 2015. Effect of cattle slurry on soil and herbage chemical properties, yield, nutrient balance and plant species composition of moderately dry Arrhenatherion grassland. Agric., Ecosyst. Environ. Appl. Soil Ecol. 213, 281–289. Duru, M., 1997. Leaf and Stem in vitro digestibility for grasses and dicotyledons of meadow plant communities in spring. J. Sci. Food Agric. 74 (2), 175–185. Elgersma, A., Søegaard, K., 2018. Changes in nutritive value and herbage yield during extended growth intervals in grass-legume mixtures: effects of species, maturity at

148