Testing and calibrating empirical models of cattle growth on native pastures in northern Australia

Abstract

Models for simulating beef-cattle enterprises require valid predictions of key animal production responses, primarily liveweight change. For the investigation of climate change or alternate management scenarios, GRASP is the most commonly-used simulation model for pasture-based cattle production in northern Australia, and is used in association with HerdEcon and Enterprise for much of the work within the Northern Grazing Systems initiative. Whilst the pasture growth components are quite detailed and widely tested, by comparison the animal production modules of GRASP are quite rudimentary. For tropical native pastures, an annual liveweight change (LWC) model in GRASP was developed using data from Mt Bambling (south-eastern Qld), Galloway Plains (central coastal Qld) and Kangaroo Hills (north-eastern Qld). The primary aim of this research is testing, extending and improving the predictions of the biological rates for beef cattle in northern Australia. Recent grazing trials across key agro-climatic regions were identified, and relevant data are currently being sourced. GRASP is to be parameterised for each of these trials, and simulated animal growth rates will be compared with actual data. In the first case-study, namely Mt Sanford in the Northern Territory, the preliminary liveweight data provided good agreement with the annual liveweight change model of GRASP. As further data sets are processed, however, it is expected that further model development and/or the identification of alternate predictive variables may be required, to address the many complexities and issues in northern Australian grazing systems such as temporal and spatial variability in the quality, quantity, utilisation and management of pasture and other vegetation, and management of animal nutrition, health and genotype. To address the issue of extrapolating the annual LWC model across landtypes and climatic zones, a wide range of expert opinions regarding 'typical' turnoff weights and rates has been obtained during the Northern Grazing Systems Project and other workshops. Compared to the number of land-types and seasonal conditions in existing data sets, these expert-opinion estimates cover a much wider range of combinations. The key coefficients fitted to these estimated weight changes appear reasonable and consistent, and also match well with those in GRASP. They can thus be used to match with existing land types, or to extrapolate to new conditions, when using GRASP for the simulation of rangeland management options. Whilst the accurate estimation of annual liveweight change is a necessary key step, in the real world tactical management decisions are typically made on a finer time-basis. These decisions include the use of forage crops, agistment or supplementation of the whole or parts of the herd, and rotational paddock grazing systems, etc. To incorporate these into an overall system model, a model for daily liveweight change is needed. The relevant logic and equations for the development of this model are listed, along with the coefficients as fitted using Solver in Microsoft Excel. Regarding liveweight changes, an acceptable degree of agreement was achieved across the base-data from native tropical pastures (887 average weights of cohorts of animals, across locations, management treatments and years). For 'annual' (> 10 month) liveweight changes the mean absolute error was 16 kg with an R 2 of 72%. Further model developments to be investigated will include the extension to sown pastures and other important beef production regions and land types. Aspects of animal performance to also be investigated include the estimation of supplementation and watermedication effects, the possible effect of age, and the extension of GRASP's 'growing steers' models to other components of beef cattle herds (e.g. heifers and cows).