Magnesium biofortification of Italian ryegrass (Lolium multiflorum L.) via agronomy and breeding as a potential way to reduce grass tetany in grazing ruminants

Aim Magnesium (Mg) deficiency (known as grass tetany) is a serious metabolic disorder that affects grazing ruminants. We tested whether Mg-fertiliser can increase Mg concentration of Italian ryegrasses (Lolium multiflorum L.) including a cultivar (cv. Bb2067; ‘Magnet’), bred to accumulate larger concentrations of Mg. Methods Under controlled environment (CE) conditions, three cultivars (cv. Bb2067, cv. Bb2068, cv. RvP) were grown in low-nutrient compost at six fertiliser rates (0–1500 μM MgCl2.6H2O). Under field conditions, the three cultivars in the CE condition and cv. Alamo were grown at two sites, and four rates of MgSO4 fertiliser application rates (0–200 kg ha−1 MgO). Multiple grass cuts were taken over two-years. Results Grass Mg concentration increased with increasing Mg-fertiliser application rates in all cultivars and conditions. Under field conditions, cv. Bb2067 had 11–73% greater grass Mg concentration and smaller forage tetany index (FTI) than other cultivars across the Mg-fertiliser application rates, sites and cuts. Grass dry matter (DM) yield of cv. Bb2067 was significantly (p < 0.05) smaller than cv. Alamo. The effect of Mg-fertiliser rate on DM yield was not significant (p ≥ 0.05). Conclusions Biofortification of grass with Mg through breeding and agronomy can improve the forage Mg concentration for grazing ruminants, even in high-growth spring grass conditions when hypomagnesaemia is most prevalent. Response to agronomic biofortification varied with cultivar, Mg-fertiliser rate, site and weather. The cost:benefit of these approaches and farmer acceptability, and the impact on cattle and sheep grazing on grasses biofortified with Mg requires further investigation. Electronic supplementary material The online version of this article (10.1007/s11104-019-04337-x) contains supplementary material, which is available to authorized users.


Introduction
Grazing landscapes and ruminant livestock have a dominant role in the environmental, economic and food security of many countries, especially in temperate regions. For example, in 2017/18, 72% of the UK's land area (17.5 M ha) was utilised for agriculture; of that area the proportion of grazing land was 35% permanent grass, 22% sole rights rough grazing, 6.9% common land rough grazing, and 6.5% temporary grass (DEFRA 2018a). These grazing lands supported a ruminant livestock population of 10 million cattle and 35 million sheep. Among EU states, the UK has the largest sheep population and the third largest cattle population, after France and Germany (DEFRA 2018b). UK agriculture contributed £10.3 billion to the national economy in 2017 with grazing ruminants (i.e., cattle and sheep) representing >50% of this total (DEFRA 2018c).
Maintaining a thriving grazing ruminant sector requires supplying livestock with balanced nutrients through forages (Agricultural Research Council 1980;McDowell and Valle 2000;Voison 1963). Magnesium (Mg) is among the essential mineral nutrients for grazing ruminants, required to ensure a healthy skeleton, metabolism, cardiovascular and neuromuscular transmission (Ebel and Günther 1980;Schonewille 2013;Underwood and Suttle 1999). Livestock dietary Mg requirement vary depending on the species, breed, physiological conditions, age and growth rate of the animal, and the type of feed (Agricultural Research Council 1980). The recommended Mg concentration for grazing ruminants ranges between 1300 and 2200 mg kg −1 DM for cattle, and 900 and 1200 mg kg −1 DM for sheep (CSIRO 2007) with the critical recommended Mg concentration of 2000 mg kg −1 DM (Mayland and Hankins 2001;McDowell and Valle 2000).
In livestock,~70% of Mg is stored in the skeleton and this pool is not easily mobilised when dietary Mg intake is reduced (Martens and Stumpff 2019;Suttle 2010). Hence, grazing ruminants need to be continuously supplied with forage that meets their Mg requirement. When the quantity of Mg supplied through feed is low, or when absorption in the rumen is impaired, the blood and cerebrospinal fluid Mg level can decline below clinical thresholds causing a physiological disorder known as hypomagnesaemic tetany (also known as grass tetany or grass staggers) (Dua and Care 1995;Henkens et al. 1973;Martens and Stumpff 2019;Suttle 2010). Ruminal absorption of Mg can be impaired by the imbalance of forage Mg 2+ , calcium (Ca 2+ ) and potassium (K + ) ions, which is termed as the forage tetany index (FTI) (Eq. 1) (Kemp and 'T hart 1957). The risk of hypomagnesaemic tetany is considered to be high in livestock consuming forage with FTI exceeding 2.2 (Crawford et al. 1998;Elliot 2008;Kemp and 'T hart 1957;Mayland and Hankins 2001;McNaught et al. 1973;Metson et al. 1966). Annually, in the UK hypomagnesaemic tetany is reported to affect 0.5% of dairy herds, and up to 10% on some dairy farms (Foster et al. 2007).
Where, mEq is the milli equivalent of the elements (i.e., the elemental concentration (mg kg −1 DM) is divided by the atomic weight and multiplied by the valence of the respective elements).
In addition to cation imbalances in the forage, lush spring grass tends to be low in fibre and high in dry matter (DM) digestibility which accelerates rumen passage and reducing the ruminal absorption of Mg (Suttle 2010). This is further exacerbated by over application of K to grasslands as fertiliser, including from livestock manures (Agricultural Research Council 1980;Bhanugopan et al. 2015;Lunnan et al. 2018). Excess K + in the soil solution suppresses the absorption of Mg 2+ by plant roots (Elliot 2008).
An Italian ryegrass (Lolium multiflorum L.) synthetic variety called Magnet (S417, Bb2067) with increased grass Mg concentration was bred by the Welsh Plant Breeding Station in the 1970s (Moseley and Baker 1991). This variety was shown to be effective in reducing hypomagnesaemic tetany in grazing sheep (Moseley and Baker 1991). However, the cultivar was never commercialised, due primarily to slightly smaller DM yield performance in National List trials in the late 1980s. The performance of cv. Bb2067 has not previously been assessed under altered Mg-fertiliser inputs. The aims of this research were (1) to test whether it is possible to raise the grass Mg concentration in Italian ryegrass by applying different rates of Mg-fertiliser, under controlled environment and field conditions; and (2) to explore the relative performance of cv. Bb2067 compared to modern Italian ryegrass cultivars under agronomic Mg biofortification.

Controlled environment (CE) experiment
The CE experiment was conducted at the Sutton All plants were sampled after 28 days, by cutting plants~1 cm above the surface. The grass was placed in paper bags and oven dried at 50°C until dry. The samples were digested in 2 mL 70% Trace Analysis Grade HNO 3 and analysed by ICP-MS as described by Thomas et al. (2016).

Field experiments design and treatments
Field experiments were conducted at Aberystwyth, Wales (52°26′00.6"N 4°00′36.7"W; 31 m.a.s.l.) and Edinburgh, Scotland (55°55′40.1"N 3°20′28.0"W; 57 m.a.s.l.) across two years (2017)(2018). The soil type at Aberystwyth is well drained loam over gravel in the Eutric Endoskeleti-Eutric Cambisols (IUSS Working Group WRB 2007) Rheidol series (Cranfield University 2019). The site at Edinburgh has a coarse textured soil (sandy-silt-loam or sandy-clay-loam) in the Macmerry series (The James Hutton Institute 2019). A randomised complete block design was adopted with four and three replications at Aberystwyth and Edinburgh, respectively. Two treatment factors (cultivar and fertiliser rate), each with four levels provided 16 treatment combinations. The cultivars were cv. Bb2067, cv. Bb2068, cv. Alamo and cv. RvP. Cultivars Bb2067 and Bb2068 are large and small Mg accumulating cultivars, respectively, which have not been commercially released. Cultivars Bb2067 and Bb2068 were bulked progeny of 2 generations of selection for Mg content. Cultivar RvP was a current variety when these selections were made and individuals with large Mg selected from this cultivar comprised~25% of cv. Bb2067 along with recurrent selections from the Lolium multiflorum breeding program. Cultivar Alamo is a modern and commonly grown commercial variety in the UK.
Plots of size 3 m * 1.2 m were sown in August 2016 at a seed rate of 35 kg ha −1 . Compound NPK fertiliser (N [22%], P 2 O 5 [4%], K 2 O [14%] SO 3 [7.5%]) was added at a rate of 60 kg ha −1 prior to the first cut (March 2017), and then at 100, 100, and 60 kg ha −1 after cuts 1, 2, 3, respectively, and then 35 kg ha −1 after all subsequent cuts. No fertiliser was added after the final cut. The Mg fertiliser was applied in April 2017 and March 2018 as magnesium sulphate (MgSO 4 ) at MgO equivalent rates of 0, 50, 100, and 200 kg ha −1 . Reagent grade ≥ 97% anhydrous MgSO 4 (Honeywell Specialty Chemicals GmbH, Seelze, Germany) was dissolved in warm water and applied with a calibrated knap sack sprayer after the first sward management cutting. Magnesium fertiliser application rates were scaled in relation to a recommendation of 50-200 kg ha −1 of MgO application every 3-4 years when exchangeable soil Mg is < 26 mg L −1 (AHDB 2017).
Grass harvesting was conducted using a Haldrup forage harvester at a cutting height of 5 cm above ground. In 2017, six grass cuts were taken at both sites. In 2018, five cuts were taken from Aberystwyth and seven from Edinburgh. Grass harvesting technique followed combined management as per UK National Lists trials protocol (Animal and Plant Health Agency 2019). During each cut, fresh weights were measured and DM yields were calculated after drying a 200-500 g subsample from each plot in a forced draught oven at 80°C for 48 h. The dried sub-sample was milled, further subsampled, digested in 2 mL 70% Trace Analysis Grade HNO 3 and analysed for concentration of Mg and other mineral elements by ICP-MS as described by Thomas et al. (2016), and certified reference materials were used to check analytical quality.
The FTI (Eq. 1) was calculated as the molar ratio of K + to the sum of Ca 2+ and Mg 2+ in the grass. Where the FTI is >2.2, the risk of grass tetany in ruminants grazing on such feed is high (Kemp and 'T hart 1957).

Soil mineral composition analyses
Prior to sowing, composite soil samples (0-15 cm depth) were collected with an auger, using a "W" transect across each site to determine baseline soil physicochemical properties. Soil samples were also collected from the 16 treatments (samples from the centre of each of four replicate plots were composited) at the beginning of June 2018, after the second Mg-fertiliser application. The baseline soil pH (in water), and exchangeable Mg, Ca and K were analysed at Lancrop Laboratory (Pocklington, UK) while the second-year soil pH, and exchangeable Mg, Ca and K concentrations were analysed at the University of Nottingham. At both laboratories, a similar procedure was followed. Thus, 5 mg of <2 mm sieved soil was dissolved in 25 mL of 1 M NH 4 NO 3. The solution was shaken on an end-overend shaker for 30 min followed by centrifuging for 15 min at 3000 rpm. The supernatant was then filtered using <0.22 μm syringe filter. The filtered solution was acidified with 0.2 mL of 50% (v/v) HNO 3 and analysed using ICP-MS (ICP-MS; iCAP-Q, Thermo-Scientific, Loughborough, UK) at the University of Nottingham (Thomas et al. 2016), and inductively coupled plasma optical emission spectrometry (ICP-OES) at Lancrop.

Data analysis
Data were compiled in MS Excel and Access. For field experiments where repeated observations (i.e., successive cuts) were made, statistical analyses were conducted using R (R Core Team 2018). Exploratory data analysis was undertaken on the residuals of an initial analysis of the data. The histogram of the residuals was inspected to assess the plausibility of an assumption of normality, and a plot of the residuals against the fitted values was inspected to assess the plausibility of an assumption that the variance of the residuals was homogeneous. When required, the data were transformed to natural logarithms to make this assumption plausible. Outliers were identified according to the outer fences of Tukey (1977) procedure whereby a datum is excluded if the residual lies more than three times the interquartile range below or above the first or third quartile, respectively. Accordingly, for grass Mg, FTI and S, the number of outlier data points excluded were two (Aberystwyth 2017), and three (Edinburgh 2018). In addition, for grass S, three more data points were excluded from Edinburgh 2018. A planned orthogonal set of contrasts was identified and mean comparison of grass Mg, FTI and S was conducted between, under field conditions, (i) the large Mgaccumulating cv. Bb2067 and other Italian ryegrass cultivars (ii) cv. Bb2068 and the two conventional varieties and (iii) cv. Alamo and cv. RvP. Under CE condition, comparison was made between i) cv. Bb2067 and the other two Italian ryegrass cultivars, and ii) cv. Bb2068 and cv. RvP.
Analyses of variances of the repeated observation (i.e., successive cuts) was addressed by the use of a linear mixed model. Two models were considered. In the first (sphericity assumption) where the correlation between the residuals for any two measures on the same unit were treated as uniform. In the second an exponential autocorrelation for successive measurements was assumed. The two models were fitted using the nlme package (Pinheiro et al. 2018) for the R platform (R Core Team 2018). The choice between the alternative models was then made based on Akaike's information criterion (AIC), selecting the model for which this was smallest.
Analysis of variance on the grass dry matter yield and soil properties in the field experiments was conducted by fitting a generalised linear model without any transformation in MINITAB 18. Visualisations were also produced using MINITAB 18 (MINITAB 2017).

Results
Raw mineral element concentration data of Italian ryegrasses of the CE experiment and field experiments, and exchangeable soil cations and dry matter yield of Italian ryegrasses data for the field experiments are given in Online Resources (Sup Table 1-4).
The total annual grass dry matter yield was affected by cultivar in both 2017 and 2018, with cv. Bb2067 and cv. Bb2068 having smaller yields than the commercial cultivars, cv. Alamo and cv. RvP. At both trial sites, the grass biomass yield in 2018 was 55-67% less than the yield in 2017 (Figs. 6 and 7) due to drought.
There were no significant (p ≥ 0.05) interaction effects of cultivar × Mg-fertiliser application rate, cultivar × cutting date, cultivar × Mg-fertiliser or cultivar × Mgfertiliser rate × cutting date on the grass biomass yield at both sites in both years. There was a highly significant (p < 0.01) negative correlation between grass biomass yield and Mg concentration. The correlation between grass DM yield and Mg concentration at Aberystwyth was (in 2017, −0.308; in 2018, −0.495), and at Edinburgh was (in 2017, −0.194;in 2018, −0.447).

Grass Sulphur (S) concentration from field experiments
At both field experimental sites, grass S concentration increased with an increasing Mg-fertiliser (MgSO 4 ) application rate for all the cultivars in 2017 and 2018 (Figs. 6 and 7). Cultivar Bb2067 accumulated significantly (p < 0.05) more S in its biomass than the other three Italian ryegrass cultivars. Planned contrasts among the cultivars mean S concentration were significant (p < 0.01) except the one between cv. Alamo and cv. RvP as shown in the Online Resource (Sup Table 6). At Aberystwyth, grass S concentration was significantly (p < 0.05) affected by cultivar, Mg-fertiliser application rate, cutting date, and by the interaction effect of cultivar × cutting date and Mg-fertiliser rate × cutting date, in 2017 and 2018 (Table 5). There were no significant interaction effects of cultivar × Mg-fertiliser application rate × cutting date on grass S concentration (Table 5). At Edinburgh, in 2017 and 2018, the grass S concentration was significantly (p < 0.05) affected by cultivar, Mgfertiliser application rate and cutting date. The cultivar × Mg-fertiliser application rate, and cultivar × cutting date interaction effect on grass S concentration was significant (p < 0.05) in 2018 but not in 2017 (Table 5). On the other hand, Mg-fertiliser rate × cutting date interaction effect on grass S was significant (p < 0.05) in both years. There was no significant (p ≥ 0.05) cultivar × Mg-fertiliser application rate × cutting date interaction effect on grass S concentration (Table 5).

Discussion
Raising grass Mg concentration to control hypomagnesaemic tetany in grazing ruminants is   Under CE conditions, we observed 85-140% increases in grass Mg concentration across three Italian ryegrass cultivars at an application rate of 1500 μM MgCl 2 . In the field, there were 7-25% increases across the four Italian ryegrass cultivars at the application rate of 200 kg ha −1 MgO equivalent of MgSO 4 . In agreement with studies elsewhere (e.g., McNaught et al. (1973) the grass Mg  These two varieties were selected as large (cv. Bb2067) and small (cv. Bb2068) Mgaccumulators originating from the same group of commercial varieties through recurrent selection. Cultivar Bb2067 clearly shows the potential for genetic biofortification. However, it had not been commercialised due to its slightly smaller dry matter yield than Italian ryegrass cultivars of its time. The herbage dry matter yield of cv. Bb2067 compared with the largest yielding cv. Alamo over two years was 86% (2017) and 76% (2018) at Aberystwyth, and 93% (2017), 85% (2018) at Edinburgh. The average annual dry matter yield cv. Alamo was reported to be 18.06 t ha −1 DM (AHDB 2018). The average dry matter yield (> 20 t ha −1 DM) of cv. Bb2067 in these trials in 2017 was well above that perennial ryegrass under conservation sward management and comparable to cv. Alamo under farmer management. There is a need to transfer cv. Bb2067 Mg-accumulating trait into the cultivars that yield larger biomass (Capstaff and Miller 2018). To facilitate this process, it might be possible to identify genetic markers that are responsible for the accumulation of Mg in cv. Bb2067.
Other potential nutritional consequences of Mg biofortification There may be additional nutrient benefits for grazing ruminants given that MgSO 4 application also increased grass S. Sulphur is an essential element for crops and livestock, and it is estimated that S deficiency is widespread in UK arable and pasturelands (Donald et al. 1999;Zhao and McGrath 1994) which could be mitigated by applying S-containing fertilisers. Here, there was no increase in DM yield due to MgSO 4 . Given the application of 7.5% SO 3 with NPK fertiliser and grass S concentration at the zero MgSO 4 application rate was >2200 mg kg −1 DM (Suttle 2010), it seems unlikely that plants were affected by S deficiency. The application of MgSO 4 to pastures can help to raise the sulphur:nitrogen (N) ratio in lush grasses with large N in spring during lambing season when animal requirement for S increases (Suttle 2010). Similarly, agronomic biofortification of forage with Mg-fertiliser can help the plants to readily take up Mg by raising the concentration of Mg 2+ in the soil solution, curbing competition from other antagonistic ions such as K + and NH 4 − , and dampening soil acidification effect on Mg 2+ availability due to frequent application of inorganic or organic Nfertiliser to pastures (Bolan et al. 2005;Mulder 1956;Voison 1963). It will be important to quantify the wider effects of using fertilisers to improve grass nutritional quality in ruminant grazing systems. For example, it would be interesting to explore if dolomitic lime (with large concentrations of Mg) were to be utilised can increase grass Ca and Mg, and reduce the FTI and help to manage pasture soil pH in those areas where this is sub-optimal.

Conclusions
This study has shown that biofortification of grass with Mg through breeding and agronomy can improve the forage Mg concentration in Italian ryegrasses for grazing ruminants, even in rapid-growth spring grass conditions when hypomagnesaemia is most prevalent. Response to agronomic biofortification varied with cultivar, Mg-fertiliser rate, site and weather. The Mg concentration in the grass biomass of the large Mgaccumulating cv. Bb2067 was greater while the FTI was smaller than the other three cultivars, at both field experimental sites. Cultivar Bb2067 consistently contained an average of >2000 mg kg −1 DM Mg at all Mg-fertiliser rates indicating its potential to reduce incidence of hypomagnesaemic tetany in grazing ruminants. Given, the slight DM yield penalty for growing cv. Bb2067 compared to cv. Alamo, transfer of the Mgaccumulating traits to the high DM yielder can be considered. The cost:benefit of these approaches, farmers' adoption, and the impact of Mg-fortified Italian ryegrasses on cattle and sheep grazing on such grasses requires further investigation.