The potential of novel Festulolium (2n=4x=28) hybrids as productive, nutrient-use-efficient fodder for ruminants

The ﬁ eld performance and potential future use of F 1 Lolium multiﬂ orum and Lolium perenne × Festuca arundinacea var. glaucescens and Festuca mairei hybrids (2n = 4x = 28) are described. Foliar trait expression in the hybrids was largely determined by the Lolium rather than their Festuca parent ensuring main-tenance of high- forage quality. All four Festulolium populations comprised high-yielding genotypes, but the L. multiﬂ orum populations were particularly erect and tall, while the L. perenne populations had signiﬁ cantly higher numbers of tillers and were prostrate. Forage yields of the Festulolium populations assessed in ﬁ eld plot trials were either not signiﬁ cantly different from, or were superior to leading L. multiﬂ orum and L. perenne cultivars used as controls. Endogenous plant proteases contribute to excessive proteolysis in the rumen which causes environmental N pollution. Protein degradation due to plant- mediated proteolysis was assessed by in vitro exposure of leaves to the environmental conditions of the rumen (39°C, anaerobic) and calculated based on the time taken for protein levels to be reduced to half their original levels ( t ½ ). Leaf proteins were signiﬁ cantly more stable in L. multiﬂ orum × F. arundinacea var. glaucescens and L. perenne × F. arundinacea var. glaucescens F 1 hybrids ( t ½ 18–21 h) than in their respective Lolium parental genotypes ( t ½ 4–5 h), and there was a highly signiﬁ cant genome interaction. The t ½ in the majority of the L. multiﬂ orum × F. arundinacea var. glaucescens F 1 hybrids studied often exceeded 24 h, whereas t ½ of their Lolium and Festuca


Introduction
In mild temperate climates such as that found in the UK ryegrass ( Lolium spp.) is often the forage grass of choice due to its high yield and nutritious value. However, recently Festulolium varieties are increasingly gaining interest as sources of reliable, productive, and nutritive fodder for use in livestock agriculture and for their potential for ecosystem services (MacLeod et al. 2013 ). Importantly, Festulolium also has a higher tolerance to

Abstract
The fi eld performance and potential future use of F 1 Lolium multifl orum and Lolium perenne × Festuca arundinacea var. glaucescens and Festuca mairei hybrids (2n = 4x = 28) are described. Foliar trait expression in the hybrids was largely determined by the Lolium rather than their Festuca parent ensuring maintenance of high-forage quality. All four Festulolium populations comprised highyielding genotypes, but the L. multifl orum populations were particularly erect and tall, while the L. perenne populations had signifi cantly higher numbers of tillers and were prostrate. Forage yields of the Festulolium populations assessed in fi eld plot trials were either not signifi cantly different from, or were superior to leading L. multifl orum and L. perenne cultivars used as controls. Endogenous plant proteases contribute to excessive proteolysis in the rumen which causes environmental N pollution. Protein degradation due to plant-mediated proteolysis was assessed by in vitro exposure of leaves to the environmental conditions of the rumen (39°C, anaerobic) and calculated based on the time taken for protein levels to be reduced to half their original levels ( t ½ ). Leaf proteins were significantly more stable in L. multifl orum × F. arundinacea var. glaucescens and L. perenne × F. arundinacea var. glaucescens F 1 hybrids ( t ½ 18-21 h) than in their respective Lolium parental genotypes ( t ½ 4-5 h), and there was a highly signifi cant genome interaction. The t ½ in the majority of the L. multifl orum × F. arundinacea var. glaucescens F 1 hybrids studied often exceeded 24 h, whereas t ½ of their Lolium and Festuca parents was consistently <14 h. Although inferior to the F 1 , F. arundinacea var. glaucescens genotypes tested had signifi cantly greater t ½ than L. perenne under rumen-simulated conditions. Signifi cant variation in protein stability was apparent within the F 1 and their respective parent species' groups. The initial protein content of the F 1 hybrids was lower than their respective parents, but following 24-h exposure to anoxia at 39°C, the protein content of both parent and hybrid genotypes was similar. The differences in protein stability between parental and hybrid genotypes was due to the greater rate of protein decline observed in the Lolium genotypes. Hence, uptake of these Festulolium hybrids as forage crops has potential to directly mitigate environmental impact of livestock farming without affecting production capacity. stresses such as drought or cold than perennial ryegrass (Ghesquière et al. 2010 ). Festulolium is the result of conventional hybridization of either Lolium perenne (perennial ryegrass) or Lolium multifl orum (Italian ryegrass) with any related Festuca (fescue) species, and may as such be marketed under its own grass category throughout Europe . Festulolium varieties may be either amphiploids with combined genome sets of ryegrass and fescue chromosomes, or they may be introgressive forms. In the latter, a limited number of donor gene sequences, most frequently derived from a fescue species, are incorporated into the recipient (ryegrass) genome through a backcross breeding program Ghesquière et al. 2010 ). The IBERS-bred variety AberNiche, the fi rst Festulolium to gain entry onto the UK National Recommended List is an example of an introgression form and is around 75% Italian ryegrass ( L. multifl orum ) and 25% meadow fescue ( Festuca pratensis ) (Cernoch and Kopecky, pers. comm.) while the French variety Lueur is an example of the amphiploid type with a more balanced ryegrass: fescue genome complement (Ghesquière et al. 2010 ). The variety Lueur derives from the hybridization of L. multifl orum (2n = 4x = 28) with F. arundinacea var. glaucescens (2n = 4x = 28), a species gaining interest due in particular to its drought and heat tolerance derived from its Mediterranean origin. The introgression breeding approach (Humphreys et al. 2005 ) has also been used to combine the complementary attributes of high-yielding ryegrass with the drought-tolerance of F. arundinacea var. glaucescens . This has been achieved through targeted markerassisted transfers of a single small genome sequence of F. arundinacea var. glaucescens onto a terminal location of chromosome 3 (Humphreys et al. 2005 ). This fescue sequence has subsequently been transferred into breeders' lines of both Italian and perennial ryegrass (Humphreys et al. 2012 ).
Lolium multifl orum × F. arundinacea var. glaucescens (2n = 4x = 28) hybrid populations have been reported to be highly palatable with voluntary intake, in vivo digestibility of organic matter (DOM), and net energy expressed in fodder units for milk, similar to some of most palatable Italian ryegrass available at that time (Ghesquière et al. 1996 ). The drought resistance of the variety Lueur derives, at least in part, from its large deep root system and its ability to extract water from depth in the soil profi le (Durand et al. 2007 ). Festuca mairei (Atlas fescue), which is related closely to F. arundinacea var. glaucescens and is indigenous to North Africa, is an alternative source of fescue genes for drought and heat tolerance and has been used (Wang and Bughrara 2005 ) as a source of novel genome variation for ryegrass through an introgression breeding approach similar to that described by Humphreys et al. ( 2005 ).
Farmers are increasingly seeking new grass varieties that provide resilience to climatic stresses, especially if these are accompanied by enhanced opportunities for environmentally sustainable livestock management. Described herein, for the fi rst time is a comparison of the agronomic potential of four amphiploid Festulolium populations: L. multifl orum × F. arundinacea var. glaucescens , L. perenne × F. arundinacea var. glaucescens (2n = 4x = 28), and L. multifl orum × F. mairei and L. perenne × F. mairei hybrids (all 2n = 4x = 28). They were assessed initially over a 1-year fi eld trial as spaced plants from which plants were selected to provide seed for a subsequent small replicated 1-year fi eld plot trial where their performance was compared with that of elite ryegrass varieties.
In addition to measures of their fi eld performance, the L. multifl orum × F. arundinacea var. glaucescens , L. perenne × F. arundinacea var. glaucescens (2n = 4x = 28), hybrid combinations were investigated for their potential, compared to ryegrass, to improve the effi ciency of ruminant nutrition. Ruminant feeds are notorious for their ineffi cient use due to the poor conversion of ingested protein to milk and meat product and as a consequence, contribute to the high greenhouse gas emissions of nitrous oxide and environmental pollution with ammonia (Ripple et al. 2014 ). In fresh forage feeding situations, rapid postingestion rates of endogenous (plant) proteolysis can contribute to rates of protein breakdown in excess of that used by microbes (Zhu et al. 1999 ;Wallace et al. 2001 ;Kingston-Smith et al. 2010 ). This occurs during autolysis and cell wall breakdown (Edwards et al. 2008 ), and through an imbalance between protein supply and energy availability for microbial growth (Johnson 1976 ). Various plant breeding initiatives have attempted to reduce gaseous emissions by livestock either by encouraging increased rates of microbial N conversion using high sugar ryegrasses (Wilkins and Humphreys 2003 ), or by delaying the degradation of plant protein in ingested feed (e.g., by stimulating polyphenol oxidase (PPO) expression, Lee et al. 2004 ), to allow more time for N assimilation by the animal. Shaw ( 2006 ) reported that F. arundinacea var. glaucescens protein was more slowly degraded under rumen conditions and had far greater protein retention than either L. perenne or L. multifl orum. Shaw ( 2006 ) speculated that this may be due to the presence of protein protective mechanisms such as heat-shock proteins that had evolved in the fescue species suffi cient to safeguard its adaptation to stresses encountered in Mediterranean conditions. Shaw ( 2006 ) also demonstrated slower breakdown of protein than its ryegrass parent in a F 1 L. multifl orum × F. arundinacea var. glaucescens (2n = 4x = 28) hybrid in rumen-simulated conditions. In order to verify and expand the fi ndings of the pilot experiments undertaken by Shaw ( 2006 ), both the same and alternative Lolium spp. × F. arundinacea var. glaucescens parental genotypes and their respective hybrids were grown under summer-like conditions in a controlled environment (CE), with plant-mediated proteolysis under the temperature and low oxygen conditions of the rumen assessed in vitro as described previously (Kingston-Smith et al., 2010) . Shaw ( 2006 ) reported the protein retention of F. mairei to be inferior to that of F. arundinacea var. glaucescens when exposed to these conditions. For this reason, the Lolium spp. × F. mairei F1 hybrids were excluded from the protein stability assessment.
Embryo rescue incorporating use of modifi ed Gamborg and Miller B5 Medium (as described in Humphreys et al. 2005 ) was employed throughout to generate all the LmFg, LpFg, LmFm, and LpFm F 1 hybrids. The LmFg populations used in the fi eld study derived from seven parental Lm genotypes of variety Danergo, six genotypes of both variety Gemini and Bb2534, and one genotype of Roberta. The Lm genotypes were hybridized onto six alternative Bn354 Fg genotypes to generate the LmFg populations. Four of these Fg genotypes were also used as parents with three genotypes of Lp variety Dunluce, and one genotype of Ba14076 and of AberDell to generate all the LpFg populations. The LmFm populations derived from three parental Lm genotypes of Bb2534, and two of variety Gemini which were hybridized onto four alternative genotypes of Fm Accession Bs3065. One of these Fm plants together with an alternative Fm genotype from Accession Bs3065 were used with two Lp genotypes of variety AberDell and two Lp cv Dunluce genotypes (also used to produce LpFg) to generate all the LpFm populations.
Genotypes of LmFg, LpFg, LmFm, and LpFm following embryo rescue were transferred to pots fi lled with potting compost and established to maturity in a frost-free glasshouse. Following growing conditions suffi cient for vernalization and infl orescence induction (>10 weeks of short days and temperatures of circa 4-10°C) and prior to ear emergence 50 F 1 LmFg genotypes, 34 F 1 LpFg genotypes, 23 F 1 LmFm genotypes, and 27 LpFm F 1 genotypes were transferred in their four groups to separate pollen-proof glasshouses to interpollinate. Seed from each polycross was germinated to produce populations for fi eld assessment as spaced plants.

Spaced-plant fi eld trial
For the fi eld study, four populations each of 300 genotypes of LmFg, LpFg, LmFm, and LpFm hybrids were established as individual spaced plants in a fi eld at IBERS, Aberystwyth University in mid-Wales, UK. Heading date was recorded in accordance with IBERS fi eld assessment protocols as number of days between 1st April and ear emergence. Aftermath heading was scored as 1 = no additional infl orescence to 9 many secondary infl orescences. Other phenotype measures were growth habit (based on a scale of 1 = erect; 5 = prostrate), plant height (cm) at ear emergence, leaf width (based on scale 1 = narrow; 9 = wide), plant width at ear emergence (cm), tiller density (based on a scale of low tiller density = 1; high tiller density = 9), disease score (based on the presence of any rust [ Puccinia spp.]) infection (score 1 = highly infected; 9 = no infection), and on their comparative plant size recorded at 5 time points throughout the growing season (score 1 = small; 9 = large). The fi rst cut was taken on 27 July 2011 with the second on the 1 September 2011.
Fifty high-yielding plants were selected from each population based on average fresh weight over the two harvests, and forage quality of this subset of plants was determined. Forage quality was measured as %water soluble carbohydrate (WSC), %dry organic matter digestibility (DOMD), and %nitrogen and total protein content.

Small-plot fi eld trial
The subsets of 50 plants were then extracted from the fi eld, repotted in potting compost and transferred to pollenproof isolation houses for seed multiplication. Seed from each LmFg, LpFg, LmFm, and LpFm polycross was sown in a second fi eld trial as three replicate fi eld plots 1 × 3 m at IBERS. Three replicate plots of control varieties Lm cv Danergo, and Gemini (4n = 4x = 28) and Lp cv AberGlyn and AstonEnergy (4n = 4x = 28) were also sown with all plots randomized. Seed of all tetraploid varieties and populations was sown at 3.3 g/m 2 (in accordance with standardized NIAB and IBERS fi eld protocols). The harvests of dry matter forage yield (DMY) were compared over six cuts taken in 2013. Forage quality: (%WSC, %dry matter digestibility (DMD) and %N and crude protein were assessed at Cuts 1, 2, and 4 using NIR technologies and in complete accordance with the standard IBERS protocols. The percentage ground cover at the end of the growing season for all grasses was compared. Detailed meteorological records for IBERS fi eld trials in 2011 and 2013 are provided in: https://share.aber.ac.uk/dept/ibers/intranet/ research/weather/default.aspx .
All fi eld data were analyzed according to standard procedures with the menu-driven options within Genstat 13.2 for Windows (VSN International Ltd. , http://www.vsni.co.uk ) software. For the spaced-plant fi eld trial, least signifi cant differences (LSD) between LpFg, LpFm, LmFg, and LmFm population means ( P < 0.05) were calculated for heading date, growth habit, plant height, leaf width, plant width, tiller density, disease score, and for plant size. For the smallplot fi eld trial, LSD ( P < 0.05) was calculated for forage yield during each of 6 cuts, and also for forage quality (%WSC, %DMD, %N) at cuts 1, 2, and 4 between the LpFg, LpFm, LmFg, and LmFm populations and their respective tetraploid Lolium control cultivars Lm cv Danergo and Gemini and Lp cv AstonEnergy and AberGlyn.

In vitro determination of plant-mediated proteolysis
The plants selected for protein analysis comprised clonal replicates of the same genotypes of Fg (4x), Lm (4x), and Lp (4x) and their respective LmFg and LpFg F 1 hybrids (all 4x) used as parents for progeny assessed in the fi eld study. They are listed in Table 1 . All plants were maintained in 6″ pots in Levington ' s multipurpose compost under identical conditions in a frost-free glasshouse at IBERS under natural illumination and watered, and when required, fertilized, cut, and repotted to encourage active and consistent plant growth throughout. Plants for protein analysis were maintained to achieve an equivalent ontogeny and with no indication of infl orescence induction to minimize potential interactions due to age difference. To further enhance consistency, all plants were transferred into a CE facility (Gallenkamp PLC Monarch Way, Belton Park, Loughborough, UK and Skye Instruments Lighting, Ddole Enterprise Park, Llandrindod Wells, Powys, UK ) and acclimated to constant UK summer conditions for a minimum of 6 weeks under 18 h light at 600 µ mol.m −2 .sec −1 at 22°C, and with 6 h darkness at 14°C, all at 72% humidity.
Mature nonsenescent leaves at equivalent developmental stage were selected from each plant genotype, removed and incubated over three concurrent days producing three replicate results. Replicate groups of parent plants and their F 1 were incubated on separate weeks due to the large number of samples. The leaves of each plant genotype were cut 5 cm above the soil into 1 cm lengths and weighed to provide equal 0.1 g fresh-weight samples for 4 incubation time points; 0, 2, 6, and 24 h. These time points were selected based on previous results (as in Shaw 2006 ). Leaf sections were transferred into Hungate tubes and fi lled with 5 mL of anaerobic buffer warmed to 39°C (Van Soest 1967 ) and, except for those at 0 h, capped quickly under a stream of CO 2 , and placed into a water bath heated to 39°C. The tubes containing incubated plant Also used previously by Shaw ( 2006 ). material were removed from the water bath after 2, 6, and 24 h, respectively. Plant material was recovered under vacuum fi ltration and rinsed with deionized water. Samples were then placed into microcentrifuge tubes, fl ash frozen in liquid N, and stored at −80°C.

Procedures for protein extraction and measurement
Batches of circa 30 incubated samples in microcentrifuge tubes were removed from the −80°C freezer, placed on ice, and transferred into racks in a freeze dryer for 24 h. The samples were then stored in the dark prior to milling. For milling, freeze-dried samples with two tungsten beads added were placed into 2 × 24 sample milling boxes and milled in a Retsch (GmbH Haan Germany) MM 300 mill at a frequency of 30 sec for 1 min on each side following rotation of the boxes. Where necessary supplementary hand grinding was applied to ensure that the samples were completely homogeneous and ground fully and uniformly. Processed samples were subsequently stored in a cool, dark, and dry place in preparation for protein extraction. Protein in the ground residue was extracted by grinding in a mortar and pestle with prechilled (4°C) extraction buffer (0.1 mol/L HEPES, pH 7.5 containing 1 mmol/L EDTA, 0.1% (v/v) Triton-X 100 and 0.5% protease inhibitor cocktail (Sigma UK Ltd, Gillingham, UK) and 2 mmol/L dithiothreitol) added to the samples at a ratio of 40 µ L/g dry weight. The sample homogenate was transferred into individual 2 mL microcentrifuge tubes, fl ash frozen in liquid N, and stored at −80°C, until protein analysis.
Homogenate samples were thawed on ice and samples centrifuged at 13,000 g at 4°C for 10 min and protein content of supernatant was determined with the Bio-Rad Protein Assay kit (Bio-Rad UK Ltd., Hemel Hempstead, UK) against a BSA calibration curve (working range 0-5 µ g). Sample volumes between 1 and 10 µ L were used to ensure results remained within range of the standard curve and water and reagent controls in a total volume of 200 µ L were included to determine background absorbance. The absorbance was read at 595 nm on a BioTek ElX 808 microtiter plate reader (Bio Tek (Bedfordshire, UK), Fisher Scientifi c ) after an incubation time of 15-20 min at room temperature.
The protein degradation time courses were fi tted with exponential decay curves of the form (1) where c is protein content (mg g −1 dry weight), t is time from start of incubation, (hours), b is a fi tted parameter describing protein content at time zero, and r is a fi tted parameter describing the rapidity of the decay of protein.
This equation was rearranged to allow calculation of the time taken in hours for protein to decay to half its value at time 0 ( t 1/2 ). (2) This parameter was chosen as the greater the t 1/2 , the greater the resistance of the plant proteins to degradation. Nonlinear curve fi tting was performed using a Maximum Likelihood Program (Ross 1987 ) and parallel curve analysis (Ross 1990 ) was used to determine signifi cant differences between fi tted curves, and to estimate standard errors of fi tted parameters and t 1/2 .

Results
Fertility in the LmFg, LpFg, LmFm, and LpFm F 1 genotypes employed as parents in their respective polycross combinations was high and more than suffi cient to provide seed for the four plant populations used in the initial fi eld study. The second seed production program all used 50 plants/population selected from the fi rst fi eld trial and as such provided an accurate comparison of population seed-set. Overall differences in seed production between the LpFm, LmFm, and LmFg populations were insignifi cant (total seed produced 277 g (LpFm), 279 g (LmFm), 271 g (LmFg), but total seed production amongst the LpFg was signifi cantly lower (seed produced = 200 g; χ 2 = 17.29, P < 0.001).

Spaced-plant fi eld trial
The mean values for the plant traits scored in the spacedplant fi eld trial and any signifi cant difference ( P < 0.05) found between the four populations are shown in Table 2 . While variation for heading date was evident within each amphiploid hybrid combination, overall no signifi cant difference was observed between the early-heading populations LpFm and LmFg (both population mean date for heading: day 37). However, populations LmFm and LpFg were both signifi cantly later heading ( P < 0.05, population mean dates for heading: day 38 and day 43, respectively). Aftermath heading was low and was not signifi cantly different in either of the Lp-based populations (LpFg and LpFm), but was signifi cantly greater ( P < 0.05) in both the Lm-based populations, especially in LmFg.
Although there were signifi cant differences ( P < 0.05) between all four populations, overall growth habit was distinct and different between the very erect Lm-based and the far more prostrate Lp-based Festulolium populations. LmFg genotypes were the most erect and were signifi cantly more erect than LmFm ( P < 0.05). Conversely, LpFg was more prostrate than LpFm ( P < 0.05) indicating that growth habit was determined more by the Lolium than the Festuca parent. LmFg was taller than LmFm ( P < 0.05). The Lm-based populations were taller than the more prostrate LpFg and LpFm populations.
There was considerable within population variation in plant size but species' effects were evident. While the Lm-based Festulolium populations were taller, the overall plant width of the Lp-based populations was greater. The mean plant width of LpFg was greater than that found in LpFm ( P < 0.05), while width of the Lm-based populations LmFg was larger than LmFm ( P < 0.05). While differences in plant habit, height, and breadth related to their Lolium parent, differences in plant mean tiller number in the Festulolium populations corresponded more to their Festuca species parent. Signifi cantly higher tiller numbers ( P < 0.05) were observed in the Fg-based populations LmFg and the LpFg, which were themselves not significantly different. The Fm-based populations LmFm and LpFm with a lower tiller frequency were not signifi cantly different.
Leaf width was determined more by the Lolium parent than by the Festuca parent, with Lm-based hybrid leaves being signifi cantly broader than those involving Lp. The mean leaf width for LmFg was signifi cantly greater than for LmFm ( P < 0.05), but both Lm-based populations had signifi cantly wider leaves ( P < 0.01) than LpFg and LpFm, which were not signifi cantly different.
Rust (predominantly Puccinia coronata ) infection was compared amongst the four amphiploid hybrid populations. Variation in disease susceptibility was evident throughout, but overall, infection was low with 90% of all plants in the fi eld trial either with low or no infection (scores 6-9). However, LmFg was more susceptible than the other three Festulolium populations and had signifi cantly higher frequencies of rust infection ( P < 0.05).
Plants in all four plant populations continued to grow and increased in plant size throughout the growing season with the 50 largest genotypes from each population selected for seed multiplication (Table 3 ). The Festulolium populations containing Fg (LmFg and LpFg) comprised larger plants than the corresponding populations with Fm (LmFm and LpFm, respectively), but the infl uence of their Lolium parent was also evident. The LmFg commenced growth and developed more extensively during the spring as compared to the other three populations, but it was the LpFg population that demonstrated most growth later in the growing season.
The dry matter yields and forage quality measures for the 50 plant LmFg, LmFm, LpFg, and LpFm selections are presented in Table 4 . The mean dry weight for cut 1 (27 July 2011) for the 50 selected plants from LmFg was signifi cantly higher ( P < 0.05) than for LmFm which in turn was signifi cantly higher ( P < 0.05) than for LpFm and LpFg. The mean dry weights from each population at cut 2 (1 September 2011) were more comparable, with LmFg and LmFm not signifi cantly different, but both were superior to LpFm and LpFg ( P < 0.05).
The %WSC of the Lp-based populations, LpFg and LpFm was not signifi cantly different at cut 1 while LmFg and LmFm were signifi cantly ( P < 0.05) lower. The superior %WSC of the Lp-based populations was maintained at cut 2 ( P < 0.05), but between the Lm-based populations, LmFm was signifi cantly higher in %WSC than LmFg. The %DMD of the Lp-based populations were not signifi cantly different and also were signifi cantly higher ( P < 0.05) than those involving Lm both at cut 1 and cut 2. The %N and total protein were also higher in the Lp-based populations over both cut 1 and cut 2 For each value within a column, populations having P < 0.05 difference are indicated by an alternative letter. ( P < 0.05). The %N and total protein of LmFg and LmFm at cut 2 was the same but at cut 1 was lower in LmFg ( P < 0.05).

Small plot fi eld trial
Spring 2013 had prolonged low temperatures delaying growth with the consequence that the fi rst harvest was delayed until 24th May. Five further cuts were made with the fi nal cut taken on 24th October. Throughout the growing season, the performance of all four Festulolium populations compared well with that of the control varieties (Fig 1 ). DMY for LmFg and LmFm and the Lm control varieties early in the year (cuts 1 and 2) did not differ signifi cantly. Likewise LpFg and LpFm although lower in yield than their Lm-based counterpart populations ( P < 0.05) did not differ for DMY from their Lp control varieties. By cut 3 LmFg had a higher yield than LmFm ( P < 0.05) but neither was signifi cantly different from the Lm control varieties. The DMY of LpFm and LpFg at cut 3 did not differ signifi cantly. The Festulolium populations did not differ in yield from the highest yielding Lp control variety (Aston Energy). At cut 4 the DMY of LmFg and LmFm and the highest yielding Lm control (variety Gemini) did not differ signifi cantly. Likewise there was no signifi cant difference in DMY between either LpFg and LpFm and the Lp control varieties. By cut 5 and 6, the forage yield of all varieties (except the inferior yielding control Lp cv AberGlyn in cut 5) irrespective of whether they were Lm-or Lp-based did not differ signifi cantly. The forage quality of the Festulolium populations harvested on Cut 1 (24th May), Cut 2 (25th June), and Cut 4 (3rd August) was consistently equivalent to or better than their respective Lm and Lp (4x) controls. The %DMD of LpFg although the highest in the fi eld trial was not signifi cantly superior to LpFm and Lp cv Aston Energy. All were superior to the Lp control AberGlyn ( P < 0.05). LmFg and LmFm had similar %DMD to the tetraploid control varieties Lm cv Danergo and cv Gemini. The %WSC of LmFm and LmFg was equivalent to Lm control cvs Danergo and Gemini. LpFm and LpFg had the same %WSC content of Lp control variety Aston Energy with all superior ( P < 0.05) to Lp cv AberGlyn. There was no signifi cant difference ( P > 0.05) in %N content and total crude protein between LpFg, LpFm, and the two Lp tetraploid cultivars Aston Energy and AberGlyn. Similarly, there was no signifi cant difference ( P > 0.05) in %N and total crude protein content between LmFg, LmFm, and the Lm controls Danergo and Gemini.
The %ground cover of the Fg-based populations LpFg and LmFg was superior to all grasses used in the fi eld trial and was signifi cantly greater than their respective controls Lp cv Aston Energy and AberGlyn and Lm cv Gemini ( P < 0.05). Figure 2 shows that the initial protein content of Lm and Fg was higher than in their F 1 hybrid progeny and differed signifi cantly ( P < 0.001). Similarly, the initial protein content of Lp and Fg was higher ( P < 0.001 for Lp, P < 0.01 for Fg), than their respective F 1 progeny. Within both the Lolium and the Festuca species' groups the initial protein content of the constituent genotypes also differed signifi cantly ( P < 0.001). There was signifi cant variation in the initial protein content between plants within the LpFg ( P < 0.05) and LmFg ( P < 0.001) F 1 hybrid groups. Despite their initial difference in protein content, following 24 h exposure to the temperature and oxygen conditions of the rumen, there remained no signifi cant difference in protein content between Lolium and Festuca parent and their respective hybrid genotypes. A comparison of t ½ of each Lm, Fg parent genotype, and their respective F1 progeny, and of Lp, Fg, and their respective F1 progeny is presented in Table 5 . In addition, and using the combined data for all time points, replicates, and genotypes for each parent and hybrid, the total t ½ values were fi tted (Fig. 3 ). Overall, the t ½ of the Fg parents was signifi cantly higher ( P < 0.05) than the Lp (4 x ) genotypes (4.5 h) used to produce the LpFg F 1 hybrids. Similarly, t ½ of the Fg parents were higher than Lm but in this case, the difference was not signifi cant ( P < 0.3). However, t 1/2 for both F 1 hybrid groups (21 h and 18 h LmFg and LpFg, respectively) were signifi cantly greater than their   respective Lolium parent groups (Lm, 4.1 h P < 0.001 and Lp 4.5 h, P < 0.01). LpFg hybrids had signifi cantly greater t ½ than their Fg parent genotypes ( P < 0.05) but the differences in t ½ between LmFg and Fg were not signifi cant ( P < 0.15). There were signifi cant differences in t ½ within the Lm ( P < 0.001), and the Lp ( P < 0.001), but not between the Fg ( P < 0.25) genotypes (Table 5 ). There were also signifi cant differences in t ½ within both the LmFg ( P < 0.001) and LpFg ( P < 0.05) hybrid groups. While collectively Fg genotypes had greater t ½ than Lp ( P < 0.05), they did not have greater t ½ than the Lm group ( P < 0.3) Within the species' groups, individual genotypes of Lm and Lp were identifi ed (Lm cv Gemini/5 and Lp cv Dunluce/5) as having superior t ½ compared to certain Fg genotypes (Bn354/17; Bn354/35). While the overall t ½ for LmFg F 1 was 21 h, genotypes with considerably higher t ½ values were identifi ed (Table 5 ). All data on protein content were confi ned to time points within 0-24 h exposures to the in vitro stress conditions and the observation in three LmFg F 1 genotypes of t ½ calculations that extended well beyond the 24 h data set led to high SE for the genotypes concerned (Table 5 ) which carried forward when all LmFg F 1 were combined (Fig 3 ).

Discussion
Breeding for Festulolium varieties has over recent years gained increased importance as an aide to combat climate change and to achieve more sustainable grassland agriculture . This is because it is possible to capture in one variety the agronomic value of ryegrass and the resilience, water, and nutrient-use-effi ciency found in different fescue species (Ghesquière et al. 2010 ). Initially, most advances and cultivars marketed involved Italian ryegrass ( L. multifl orum ) and meadow fescue ( F. pratensis ), a species combination known taxonomically as Festulolium braunii. However, in recent years especially at IBERS attention has moved more to employing alternative fescue species such as F. arundinacea var. glaucescens and more recently F. mairei . These have provided novel sources of genes for improved drought resistance, water-use-effi ciency, and deep rooting for ryegrass (Humphreys et al. 2005. The plant-soil interactions generated by certain deep rooting Festulolium species combinations including those described herein and others such as L. perenne × F. pratensis ( Festulolium loliaceum) can provide other benefi ts including ecosystem services such as fl ood mitigation MacLeod et al. 2013 ) and have potential to increase soil organic carbon capture by grassland (Kell 2011 ) . While the potential of Festulolium to improve grassland persistency has long been recognized, for reasons such as genome instability (Canter et al. 1999 ), high costs of seed production compared to ryegrass, and inferior forage quality (Ghesquière et al. 2010 ) Festulolium has not yet been widely used. For these reasons, doubts surrounding its commercial development remain and these limitations must be overcome before Festulolium varieties become marketed widely and their full benefi ts realized. However, recent advances in Festulolium breeding technologies combined with an increased awareness of the need to fi nd alternatives to ryegrass to better combat climate change and to Table 5 . Protein half-lives ( t ½ ) (SE) of (i) Lolium multifl orum (Lm), Festulolium arundinacea var. glaucescens (Fg), and their F1 progeny (all 2n = 4x = 28), and (ii) L. perenne (4x), F. arundinacea var. glaucescens (Fg), and their F1 progeny (all 2n = 4x = 28) when exposed in vitro to 24 h of anoxia at 39°C. achieve more sustainable grassland systems have encouraged Festulolium development and led to the marketing of the cultivar AberNiche in the UK, an expected forerunner for others under development and trial. AberNiche is a synthetic form of Festulolium braunii comprising genome combinations of L. multifl orum and F. pratensis . It was exluded from the current study as in earlier work (Shaw, 2005) , F. pratensis demonstrated a similar plant-mediated proteolysis to that recorded for Lolium spp., and was signifi cantly inferior to the protein retention found in F. arundinacea var. glaucescens when exposed to the stress conditions applied in the current study. In an alternative fi eld trial at IBERS AberNiche, L. multifl orum × F. arundinacea var. glaucescens (LmFg) , and L. multifl orum × F. mairei (LmFm), all Lm-based tetraploid Festulolium hybrids, produced an equivalent forage yield throughout their fi rst year harvests (Humphreys, unpubl. (Ghesquière et al. 1996(Ghesquière et al. , 2010 but in trials at IBERS has inferior yield to LmFg (Humphreys, unpubl.).
Two possible strategies for increasing effi ciency of conversion of forage N to microbial N have been used to improve ruminant nutrition and decrease emissions of greenhouse gases by livestock and both have received considerable recent attention at IBERS over recent years. The fi rst strategy aims at increasing the amount of readily accessible energy during the early part of the fermentation in the rumen. The second strategy aims to increase the protection of forage proteins, and thereby reducing the rate at which their breakdown products are made available to the colonizing microbial population. The high sugar grasses (HSG) are examples of the former where increased WSC has been shown to have a positive impact on meat yields (Lee et al. 2001 ) and milk production (Miller et al. 2001 ). The incorporation in legumes of protein protection methods such as increased PPO expression is an example of the second approach as applied to conserved forage (Lee et al. 2004 ) while increasing the t 1/2 for protein degradation in the rumen would increase N use effi ciency by grazing ruminants, as discussed below.
Extensive nitrogen loss to the environment due to poor incorporation of dietary N by ruminants causes both pollution of ground water and contributes to nitrous oxide production. This is due to excessive proteolysis in the rumen, for which plant-mediated proteolysis is a contributory factor (Zhu et al. 1999 ;Wallace et al. 2001 ;Kingston-Smith et al. 2005 ) and hence current efforts to mitigate impact of ruminant farming through selection of improved forage genotypes. Shaw ( 2006 ) demonstrated that Fg was signifi cantly more stable under rumen-like conditions than Lolium species. She also demonstrated that Fg was more stable than Fm when exposed to equivalent rumensimulated trials. Although Shaw ( 2006 ) reported benefi ts in terms of improved protein stability in F 1 hybrids compared with Lm, these were lost during the course of a backcross breeding program identical to that described in Morgan et al. ( 2001 ). The current work sought to extend and verify the study described in Shaw ( 2006 ) by exploring the variation within Lp, Lm, and Fg for protein stability under rumen-like conditions and to determine the extent at which this trait is expressed in Festulolium F1 hybrids. Evidence is presented here to show (i) signifi cant variation for protein stability within Lm, Lp, and Fg tetraploid genotypes, (ii) that the range of protein stability in Lm, Lp, and Fg genotypes has signifi cant overlap, (iii) signifi cant Lolium-Festuca genome interactions and transgressive segregation in F 1 hybrids between Fg and both Lm and Lp that gave rise to signifi cantly higher protein stability than that expressed by either Lolium parent genomes. Shaw ( 2006 ) proposed that protein-protection mechanisms that had evolved in Fg to combat the high temperatures experienced in Mediterranean locations were providing equivalent benefi ts when grass was exposed to the stresses encountered in the rumen. The current work indicates that the genetic control for protein stability is complex and demonstrates signifi cant variation within and overlapping protein stability between Lm, Lp, and Fg populations. Heterosis between the Lolium and Festuca genomes in their F 1 hybrid forms provided improved protein stability compared with the Lolium and Fg parents. The initial protein content of the Lp, Lm, and Fg parent genotypes was consistently higher than that observed in their progeny. However, following exposure to 24 h of rumen-simulated conditions, despite signifi cant difference in their protein half-lives, there was little or no actual difference in residual protein content between parent and hybrid genotypes. A similar relationship between initial low protein content and slow rates of protein degradation was seen previously with white clover (Kingston-Smith et al. 2006 ). While the outcome in terms of protein content in both Lolium and the Festulolium hybrid genotypes at 24 h was similar, the rate of protein decline was signifi cantly greater in the Lolium genotypes. Protein degradation in the early time period following ingestion of forage is considered to be important. Despite the availability of peptide and amino acid substrates for microbial growth at this stage, the availability of energy will be determined (and possibly limited) by the extent of microbial colonization and cell wall degradation (Johnson 1976 ;Edwards et al. 2008 ). Hence, decreasing plant-mediated proteolysis has the potential to improve delivery of protein and energy by the feed. As protein building blocks are nonlimiting the main consequences of decreasing plant-mediated proteolysis would be predicted to be decreased activity of HAP bacteria and, on a whole animal level, increased N partitioning to product and away from urine. It will be important to now extend the current work by employing animal studies to fully assess the benefi ts of LmFg and LpFg hybrids as feed both in terms of their potential for improved ruminant nitrogen-use-effi ciency and livestock gain and also for environmental gain by limiting N losses into the environment.
In fi eld trials, genotypes of four Festulolium populations; LmFg, LpFg, LmFm, and LpFm were selected for seed multiplication; seed set from these and subsequent selections was far in excess of that required for the initial spaced plant trial and subsequent plot trial experiments. The two Lm-based populations were large and erect and contrasted with the two Lp-based populations which were large and high tillering and consistently prostrate. For the majority of foliar traits, it was the Lolium (Lm or Lp) rather than the Festuca (Fg or Fm) parent that was the determining factor in trait expression as reported previously in a root phenotype study  ). In the earlier work, root ontogeny was found to be more dependent on the presence of an Lm or an Lp genome rather than whether the accompanying genome complement was Fg or Fm. In the fi eld plot trial, all four Festulolium populations compared favorably and were not signifi cantly different in yield or forage quality from their respective Lm and Lp (4 x ) control varieties demonstrating the absence of any suggestion of transfers of deleterious forage characters from their fescue parent.
The combined fi eld-based and in vitro proteolysis study provides evidence that both LpFg and LmFg amphiploid hybrids offer considerable potential for sustainable grassland agriculture. The LmFm and LpFm populations provided similar benefi ts and would be expected to be particularly drought and heat tolerant (Wang and Bughrara 2005 ). The root systems of all four populations combined the high growth rate and branching of Lolium with the root strength and depth of Fg and Fm (data not shown, Humphreys et al. 2013 ). The impact of these Festulolium hybrids in terms of plant-soil interactions is being assessed currently and will be compared with earlier research (MacLeod et al. 2013 ) for potential ecosystem service benefi ts. Taken together, the outcomes of the current research provide compelling evidence for benefi ts both for agriculture and the environment for future use of Festulolium hybrids. nutrition are presented here. We also acknowledge BBSRC for funding the research inputs from Alison Kingston-Smith, Mike Humphreys, and Alan Gay. The IBERS fi eld trials and seed multiplication were funded through an industrial DefraLINK Consortium (DefraLink Programme LK0688) led by Mike Humphreys and undertaken by Markku Farrell.

Confl ict of Interest
None declared.