Modelling the consequences of targeted selective treatment strategies on performance and emergence of anthelmintic resistance amongst grazing calves
The development of anthelmintic resistance by helminths can be slowed by maintaining refugia on pasture or in untreated hosts. Targeted selective treatments (TST) may achieve this through the treatment only of individuals that would benefit most from anthelmintic, according to certain criteria. However TST consequences on cattle are uncertain, mainly due to difficulties of comparison between alternative strategies. We developed a mathematical model to compare: 1) the most 'beneficial' indicator for treatment selection and 2) the method of selection of calves exposed to Ostertagia ostertagi, i.e. treating a fixed percentage of the population with the lowest (or highest) indicator values versus treating individuals who exceed (or are below) a given indicator threshold. The indicators evaluated were average daily gain (ADG), faecal egg counts (FEC), plasma pepsinogen, combined FEC and plasma pepsinogen, versus random selection of individuals. Treatment success was assessed in terms of benefit per R (BPR), the ratio of average benefit in weight gain to change in frequency of resistance alleles R (relative to an untreated population). The optimal indicator in terms of BPR for fixed percentages of calves treated was plasma pepsinogen and the worst ADG; in the latter case treatment was applied to some individuals who were not in need of treatment. The reverse was found when calves were treated according to threshold criteria, with ADG being the best target indicator for treatment. This was also the most beneficial strategy overall, with a significantly higher BPR value than any other strategy, but its degree of success depended on the chosen threshold of the indicator. The study shows strong support for TST, with all strategies showing improvements on calves treated selectively, compared with whole-herd treatment at 3, 8, 13 weeks post-turnout. The developed model appeared capable of assessing the consequences of other TST strategies on calf populations.
Evaluating the feasibility of developing a model to better manage nematode infections of sheep
This study evaluates the feasibility of developing (or accessing) a sheep nematode epidemiology model for Australian conditions. Following consultation with animal health experts, such a model would need to predict the impact of integrated parasite control strategies (nutrition, grazing management, anthelmintic treatment strategies and selective breeding for resistance) upon productive traits, parasitological traits and the emergence of anthelmintic resistance. Seven existing nematode epidemiology models were reviewed to evaluate their suitability for Australian conditions in their current form, or after customisation. Whilst individually these models were found to be incapable of evaluating integrated parasite control strategies, a composite of these models could achieve this aim. The best functions from the models reviewed were identified and the initial outline of a composite model is consequently proposed. Access to such a model for industry advice, educational or research purposes can be facilitated via its inclusion in the WormBoss website following development of a user friendly interface. Further, providing open-access to the model source code will inform researchers of underlying assumptions, allow for thorough review, remove reliance upon an individual, and facilitate further development. Finally, the potential pathway and cost of developing a validated sheep nematode epidemiology model and advice tool is considered.
Which is the best phenotypic trait for use in a targeted selective treatment strategy for growing lambs in temperate climates?
Targeted selective treatment (TST) requires the ability to identify the animals for which anthelmintic treatment will result in the greatest benefit to the entire flock. Various phenotypic traits have previously been suggested as determinant criteria for TST; however, the weight gain benefit and impact on anthelmintic efficacy for each determinant criterion is expected to be dependent upon the level of nematode challenge and the timing of anthelmintic treatment. A mathematical model was used to simulate a population of 10,000 parasitologically naïve Scottish Blackface lambs (with heritable variation in host-parasite interactions) grazing on medium-quality pasture (grazing density = 30 lambs/ha, crude protein = 140 g/kg DM, metabolisable energy = 10 MJ/kg DM) with an initial larval contamination of 1000, 3000 or 5000 Teladorsagia circumcincta L₃/kg DM. Anthelmintic drenches were administered to 0, 50 or 100% of the population on a single occasion. The day of anthelmintic treatment was independently modelled for every day within the 121 day simulation. Where TST scenarios were simulated (50% treated), lambs were either chosen by random selection or according to highest faecal egg count (FEC, eggs/g DM faeces), lowest live weight (LW, kg) or lowest growth rate (kg/day). Average lamb empty body weight (kg) and the resistance (R) allele frequency amongst the parasite population on pasture were recorded at slaughter (day 121) for each scenario. Average weight gain benefit and increase in R allele frequency for each determinant criterion, level of initial larval contamination and day of anthelmintic treatment were calculated by comparison to a non-treated population. Determinant criteria were evaluated according to average weight gain benefit divided by increase in R allele frequency to determine the benefit per R. Whilst positive phenotypic correlations were predicted between worm burden and FEC; using LW as the determinant criterion provided the greatest benefit per R for all levels of initial larval contamination and day of anthelmintic treatment. Hence, LW was identified as the best determinant criterion for use in a TST regime. This study supports the use of TST strategies as benefit per R predictions for all determinant criteria were greater than those predicted for the 100% treatment group, representing an increased longterm productive benefit resulting from the maintenance of anthelmintic efficacy. Whilst not included in this study, the model could be extended to consider other parasite species and host breed parameters, variation in climatic influences on larval availability and grass growth, repeated anthelmintic treatments and variable proportional flock treatments.
Modelling the short- and long-term impacts of drenching frequency and targeted selective treatment on the performance of grazing lambs and the emergence of anthelmintic resistance
Refugia-based treatment strategies aim to prolong anthelmintic efficacy by maintaining a parasite population unexposed to anthelmintics. Targeted selective treatment (TST) achieves this by treating only animals that will benefit most from treatment, using a determinant criterion (DC). We developed a mathematical model to compare various traits proposed as DC, and investigate impacts of TST and drenching frequency on sheep performance and anthelmintic resistance. Short term, decreasing the proportion of animals drenched reduced benefits of anthelmintic treatment, assessed by empty body weight (EBW), but decreased the rate of anthelmintic resistance development; each consecutive drenching had a reduced impact on average EBW and an increased impact on the rate of anthelmintic resistance emergences. The optimal DC was fecal egg count, maintaining the highest average EBW when reducing the proportion of animals drenched. Long-term, reducing the proportion of animals drenched had little impact on total weight gain benefits, across animals and years, whilst reducing drenching frequency increased it. Decreasing the frequency and proportion of animals drenched were both predicted to increase the duration of anthelmintic efficacy but reduce the total number of drenches administered before resistance was observed. TST and frequency of drenching may lead to different benefits in the short versus long term.
A stochastic model to investigate the effects of control strategies on calves exposed to 'Ostertagia ostertagi'
Predicting the effectiveness of parasite control strategies requires accounting for the responses of individual hosts and the epidemiology of parasite supra- and infra-populations. The first objective was to develop a stochastic model that predicted the parasitological interactions within a group of first season grazing calves challenged by 'Ostertagia ostertagi', by considering phenotypic variation amongst the calves and variation in parasite infra-population. Model behaviour was assessed using variations in parasite supra-population and calf stocking rate. The model showed the initial pasture infection level to have little impact on parasitological output traits, such as worm burdens and FEC, or overall performance of calves, whereas increasing stocking rate had a disproportionately large effect on both parasitological and performance traits. Model predictions were compared with published data taken from experiments on common control strategies, such as reducing stocking rates, the 'dose and move' strategy and strategic treatment with anthelmintic at specific times. Model predictions showed in most cases reasonable agreement with observations, supporting model robustness. The stochastic model developed is flexible, with the potential to predict the consequences of other nematode control strategies, such as targeted selective treatments on groups of grazing calves.
In silico exploration of the mechanisms that underlie parasite-induced anorexia in sheep
A model was used to investigate two mechanisms describing reductions in food intake (anorexia) observed during gastrointestinal parasitism in lambs, and to explore relationships between anorexia and food composition. The mechanisms were either a reduction in intrinsic growth rate, leading to a consequent reduction in food intake (mechanism 1; M1), or a direct reduction in food intake (mechanism 2; M2). For both mechanisms, lambs growing from 2 to 6 months of age were modelled, with one of three levels of trickle challenge with 'Teladorsagia circumcincta'. Scenarios were simulated for feeds varying in either protein or energy content, or both. Major differences were found between the predictions resulting from M1 and M2 on low-energy foods that constrained the intake of uninfected lambs through bulk. With M1, food intake was governed by the first operating constraint, whereas with M2 an additivity of constraints was observed. On the other foods, the duration of anorexia increased with increasing energy content of feed for M1, whilst the duration of anorexia decreased with increasing protein content of feed for M2. For foods that did not have an impact upon lambs' gastrointestinal tract capacity, published data were consistent with predictions of M2. Due to an absence of experimental data, no conclusions could be drawn for relationships between anorexia and food composition in the presence of other limiting constraints, such as bulk for low-energy foods. In conclusion, available experimental data and model predictions were consistent with anorexia having an impact directly on food intake, and with impacts of anorexia increasing with decreasing protein content.
Exploration of the epidemiological consequences of resistance to gastro-intestinal parasitism and grazing management of sheep through a mathematical model
Predicting the impacts of selection for decreased faecal egg count (FEC) (i.e. host resistance) in grazing ruminants is difficult, due to complex interactions between parasite epidemiology, management and host responses. A mathematical model including heritable between lamb variation in host-parasite interactions, 'Teladorsagia circumcincta' epidemiology and anthelmintic drenching, was developed and used to (i) address such interactions and their impact on outcomes including FEC, live weight (LW, kg) and pasture larval contamination (PC, larvae/kg DM), and (ii) investigate how grazing management strategies, aimed at reducing host exposure to infective larvae via pasture moves at 40 day intervals, affectthese outcomes. A population of 10,000 lambs was simulated and resultant FEC predictions used to assign the 1,000 lambs with the highest and lowest predicted FEC to 'susceptible' (S) and 'resistant', (R) groups, respectively. The predicted average FEC of the S group was ∼8.5-fold higher than the R group across a grazing season. The R and S groups were then simulated to graze separate pastures (Rsep and Ssep); and repeated for 3 grazing seasons to allow predictions to diverge and stabilize. Further, different grazing strategies were superimposed on all groups. PC and average FEC were affected by whether lambs of different resistance genotype grazed together or separately, with differences increasing across grazing seasons. By the third grazing season the average PC of the Rsep group was reduced by ∼83%, and the Ssep group was increased by ∼240%, in comparison to the whole population average. Average FEC of the Rsep group was reduced by ∼40%, and the Ssep group increased by ∼46% in comparison to the R and S groups, respectively, whilst drenching had little impact on the proportional differences in FEC between groups. Predicted LW was similar for the R and Rsep groups irrespective of anthelmintic treatment, whilst LW of the Ssep group was reduced by ∼14% compared to the S group for un-drenched lambs, and by ∼4% for drenched lambs. The differing grazing strategies were predicted to have little impact on FEC or LW, with the exception of the Ssep group which was predicted to have a 2 kg increase in LW when drenched and moved to a clean pasture. Together, these results suggest that host genotype has a substantial impact on parasite epidemiology, however the benefits of anthelmintic treatment and grazing management should only be expected for susceptible animals. This supports the use of targeted selective treatment, focussing on susceptible animals.
Can we use genetic and genomic approaches to identify candidate animals for targeted selective treatment
Estimated breeding values (EBV) for faecal egg count (FEC) and genetic markers for host resistance to nematodes may be used to identify resistant animals for selective breeding programmes. Similarly, targeted selective treatment (TST) requires the ability to identify the animals that will benefit most from anthelmintic treatment. A mathematical model was used to combine the concepts and evaluate the potential of using genetic-based methods to identify animals for a TST regime. EBVs obtained by genomic prediction were predicted to be the best determinant criterion for TST in terms of the impact on average empty body weight and average FEC, whereas pedigree-based EBVs for FEC were predicted to be marginally worse than using phenotypic FEC as a determinant criterion. Whilst each method has financial implications, if the identification of host resistance is incorporated into a wider genomic selection indices or selective breeding programmes, then genetic or genomic information may be plausibly included in TST regimes
WormLoad: a pasture infectivity risk model of four nematode species in Australia
Gastrointestinal parasites cost the Australian sheep industry AU$436 million annually. Early warning of impending worm risk may reduce this cost by providing producers sufficient time to implement control strategies. The provision of 90 day weather forecast data at a 6km grid resolution across Australia (Australian Bureau of Meteorology) has enabled the development of a mathematical model to predict the risk arising from nematode pasture infectivity for inclusion in the Sheep CRC's 'AskBill' application. A biophysical modelling approach was used to simulate the on-pasture lifecycle stages of 4 nematode species (Haemonchus contortus, Teladorsagia circumcincta, Trichostrongylus colubriformis and Trichostrongylus vitrinus). Mortality and development/migration rates of each lifecycle stage were described using modified β-distribution functions to account for the impact of temperature and water availability. The model was parameterised against point estimates from available literature and experimental data for the 4 species (H.contortus: R 2 = 0.88, n = 1409; T.circumcincta: R 2 = 0.56, n = 243; T.colubriformis: R 2 = 0.61, n = 355; T.vitrinus: R 2 = 0.66, n = 147). In the absence of a model predicting the quantity of eggs deposited, a probabilistic approach was used assuming 1 egg species-1 sheep-1 day-1 (thereby accounting for stocking rate). The impact of anthelmintic treatments were accounted for by assuming that no eggs were deposited for the duration of claimed efficacy. Risk was calculated by summing the proportion of infective larvae available for ingestion for each nematode species across each day of prior egg deposition and adjusting for herbage availability, species fecundity and productive impact.
A mathematical model to predict the risk arising from the pasture infectivity of four nematode species in Australia
Gastrointestinal parasites cost the Australian sheep industry AU$436 million annually. Early warning of impending worm risk may reduce this cost by providing producers with sufficient time to implement control strategies. A biophysical model was developed to simulate the on-pasture lifecycle stages of the four predominant nematode species in Australia (Haemonchus contortus, Teladorsagia circumcincta, Trichostrongylus colubriformis and Trichostrongylus vitrinus). The influence of climatic variables (temperature and water availability) on the survival, development and migration of each lifecycle stage was incorporated and parameterised to available point estimates (H. contortus: R² = 0.88, n = 1409; T. circumcincta: R² = 0.56, n = 243; T. colubriformis: R² = 0.61, n = 355; T. vitrinus: R² = 0.66, n = 147). Constant fecundities (eggs/worm.day) provided the daily quantity of eggs deposited per sheep (H. contortus = 3275; T. circumcincta = 140; T. colubriformis = 300; T. vitrinus = 300). Farm management practices were considered via the specification of stocking rates (sheep/ha), and the administration of anthelmintic treatments (reducing egg deposition by a defined efficacy and duration for each nematode species). Pasture infectivity per nematode species was calculated as the quotient of larvae on herbage and herbage availability (t/ha). Risk was calculated as the product of pasture infectivity and the potential productive impact of each nematode species (H. contortus = 3.9%; T. circumcincta = 9.22%; T. colubriformis = 9.31%; T. vitrinus = 9.31%), and then summed across nematode species. This predictive model has been incorporated into the Sheep CRC's 'ASKBILL' application (www.askbill.com.au, verified 13 April 2018), which uses 90-day weather forecast data (5-km grid resolution) provided by the Australian Bureau of Meteorology.