Ryegrass endophyte mixtures for improved animal health : A thesis submitted in partial fulfilment of the requirements for the Degree of Doctor of Philosophy at Lincoln University

In New Zealand, perennial ryegrass (Lolium perenne) with endophyte (Epichloë festucae var. lolii) is a standard ingredient of pastures where insect pests challenge the persistence of the ryegrass. Farmers currently have multiple ryegrass cultivar x endophyte strain combinations from which to choose the best match for their requirements. However, no single cultivar x strain combination provides both the highest insect protection and lowest occurrence of ryegrass staggers and heat stress in animals grazing pasture with endophyte. Furthermore, the current industry protocol for testing the safety of animals grazing a grass x endophyte combination involves exposing the animals to a simulated worst- case scenario and needs reassessment. This research aimed to examine the effects of endophyte diversity in perennial ryegrass on the health and performance of sheep using a modelling approach so that an animal response of any mixture could be predicted as a function of the mixture’s endophyte proportions, thus minimising the need for animal testing in the future. At Barenbrug Plant Breeding Station, Courtenay, Canterbury, eight pasture treatments of Maxsyn perennial ryegrass – seven that varied widely in sown proportions of three endophyte strains – nea3, nea12, and standard endophyte (SE) – and an endophyte-free control (nil), were tested over four experimental runs from February 2020 to December 2021. There were three monocultures (100% of each strain), three binary mixtures (50% of each of two strains) and one ternary mixture (33.3% of each strain) of endophyte based on a simplex centroid design. The eight treatments were randomised in three blocks of 0.156-ha plots. The animal responses of interest were ryegrass staggers, liveweight gain, heat stress, and diet selection of grazing sheep. The first experiment run was a test for ryegrass staggers in February 2020. Endophyte toxicity in late summer is typically near its annual maximum in Canterbury, so the timing of this run created a ‘worst-case scenario’ for ryegrass staggers. The pastures were managed in the previous spring so as to accumulate a large herbage dry matter (DM) mass (about 3700 kg DM/ha above a cutting height of 40–50 mm) of low nutritive quality feed (metabolisable energy (ME) = 8.7 MJ/kg DM and crude protein (CP) = 4.9%). The pasture available (kg DM/ha) was high enough for lambs to remain in the same plots for a 4-week testing period at a stocking rate of 64.1 lambs/ha. Ryegrass staggers was severe (score 4–5) in the lambs grazing the SE pasture after 8 days and in all other endophyte monocultures and mixtures after 22 days. Mixture models fitted to the staggers response data predicted the staggers score for any combination of sown endophyte proportions and identified an optimum of 65% nea3 and 35% nea12 which delayed severe staggers by 1 week. Ryegrass staggers was associated with tremorgenic alkaloids in the herbage above 40–50 mm: epoxy-janthitrem I of nea12 (1 ppm), paxilline and terpendole C of nea3 (0.1 and 1.2 ppm), and lolitrem B, paxilline and terpendole C of SE (2.2, 0.2, and 1.1 ppm). The second experimental run was a study of liveweight gain and ryegrass staggers under low toxicity conditions from September to December 2020. The spring pastures provided optimal conditions for liveweight gain and the stocking rate was kept consistent at 44.9 hoggets/ha. Pre-grazing pasture mass was ca. 1600 kg DM/ha, ME was 12 MJ/kg DM and CP was 20% above 40–50 mm. The average daily gain for all mixtures was 271 g/day over the first 4 weeks, and 20 g/day for weeks 4–8. The mixture models fitted to the liveweight gain data identified the optimum endophyte formulation that maximised liveweight gain to be a monoculture of nea12 so that the responses were equal to the monoculture performance in the first and second 4 weeks of grazing (292 and 25 g/day). Ergovaline, the alkaloid linked to suppressed liveweight, was present in nea3 and SE pastures at 0.32 and 0.16 ppm. Staggers scoring at the onset of symptoms after 10 weeks of grazing revealed mild staggers (score <2) in SE that persisted until the experiment ended at 12 weeks. The model analysis showed that any mixture containing at least 60% nea3 and up to 40% nea12 and/or SE would result in no staggers (score 0). Alkaloids associated with staggers were in lower concentrations compared to the summer test: 0.8 ppm of epoxy-janthitrem I was present in nea12 pastures while lolitrem B, paxilline, and Terpendole C were present in SE pastures at 1, 0.1, and 0.3 ppm. The third experimental run tested the assumption that endophyte strain proportions in the diet were equal to the sown proportions. Plot fences within replicates were removed, 51–56 hoggets were allocated to each replicate (40.9–44.9 hoggets/ha), and sward height decrease was measured 15 times between 15 February and 16 March 2021. Herbage mass, ME, and CP values at the beginning of grazing were circa 1400 kg DM/ha, 11.5 MJ/kg DM, and 11%. Analysis of variance showed no differences (P>0.05) in sward height decrease between sown treatments, including nil, over the whole grazing period. Mixture models of each measurement date and the average of all dates only showed differences (P<0.05) between linear terms (monocultures) on day 2 and further analysis confirmed the sward height decrease of SE was greater than nea12 (P<0.05), but not nea3. Quadratic terms (binary mixtures) were not different on any date. The mixture model analysis for the whole period predicted that a mixture containing 39% nea3 and 61% SE would maximise sward height decline despite lolitrem B concentrations of SE pastures being sufficient to cause ryegrass staggers (1.4 ppm) and ergovaline levels in nea3 pastures of 0.37 ppm. The lack of deterrence to these alkaloids indicates that animals were unable to differentiate between endophyte strains. The fourth experimental run was a repeat liveweight gain test followed by a heat stress study from October to December 2021. Pasture conditions were similar to the previous liveweight gain test at the start of grazing (ca. 2400 kg DM/ha, 12 MJ ME/kg DM, 15% CP). The same number of hoggets were used as in the previous liveweight gain study, but a higher starting liveweight and greater pre- grazing pasture mass resulted in greater liveweight gain across mixtures in the first and second 4 weeks (397 and 67 g/day) compared to the previous spring. The model analysis predicted an optimum mixture containing 16% nea3, 65% nea12, and 20% SE would result in maximum liveweight gains of 402 and 85 g/day in the first and second 4 weeks of the test. Data loggers attached to controlled internal drug release (CIDR) devices recorded vaginal temperatures in selected mixtures (nea3, nea12, nea3–nea12, and nil) during the final 8 days of the 8-week grazing period. Average temperature for the three mixtures was 39.07°C compared to 38.96°C in the nil endophyte control. Hoggets grazing nea12 had the lowest temperature of the three mixtures (38.94°C; P<0.05). Consequently, the mixture model predicted the optimum mixture would only contain nea12. Ergovaline concentrations were 0.7 and 0.5 ppm in nea3 and SE. An endophyte diversity model that included ryegrass staggers and liveweight gain responses from experimental runs 1, 2, and 4 as multivariates, predicted a mixture containing 77% nea12 and 23% nea3 would give the best overall combination of responses. Although temperature was not included in the model, the vaginal temperature response of this optimal mixture was predicted to be 39.02°C, just 0.06°C higher than the nil endophyte control. This thesis showed that endophyte mixtures can improve the health of grazing sheep and provides the industry with a new option for mitigating health issues associated with endophyte without searching for novel strains. Such mixtures would be best suited to scenarios where insect pest pressure requires the use of endophyte strains that can increase the risk of ryegrass staggers and/or suppressed liveweight gain. The endophyte diversity modelling approach can predict responses to sown strain proportions beyond those included in the experiment, reducing the number of animal entries required for testing.