Effects of grazing management and pasture composition on the nitrogen dynamics of a dairy farm: a simulation analysis

Abstract

There is an extensive debate on the potential environmental impact of dairy farms and in particular the effect of dairy farms on the nitrogen cycle and the effect that this has on ecosystems. Within New Zealand and in particular in the South Island, the expansion of dairying and the adoption of new dairy systems has led to this becoming an increasingly important issue, locally through its effect on water quality and the environment and nationally and internationally through the production of green house gases. Increases in nitrogen usage at the expense of clover nitrogen fixation, irrigation, stocking rate and the introduction of dairy cows onto light free draining soils previously the preserve of arable or sheep farming has led to concern as to the effect intensive pastoral dairying may have on the nitrogen dynamics of the farm and the environment. This study is designed to assess how changes in grazing management in particular changes in pre-grazing and post-grazing residuals alter the clover/ryegrass balance on the farm and the effect that this has on the farm’s nitrogen dynamics. The effects of qualitative changes in grazing management on pasture composition are well established but little is known of the effect of quantitative changes in pasture management on composition, in particular the effect of grazing residuals. There are a number of detailed models of the physiological processes in the energy and nutrient cycling in plants, animals and the soil. There are a smaller number of whole farm models that through integration and simplification of component models attempt to represent the flux of nutrients though a dairy farm. None of these whole farm models is currently able to model the nitrogen flux through a dairy farm at a sufficient level of resolution to capture differences in pasture composition as these occur spatially, temporally and in response to grazing management. This project sought to better understand the nitrogen dynamics on a dairy farm by constructing and then linking component models – a pasture composition and growth model, a cow model, an excretal return model, a soil model and a water balance model – within a whole farm management schedule. The formal null hypothesis is that the mechanistic, mathematical model constructed for this PhD cannot capture and explain the full range of the changes in soil water content, soil nitrogen status, pasture production and composition and animal production, following the alteration in management of the dairy farm between 2002 and 2004. Individual component models were constructed by the author using the computer software package (Matlab) and validated against data extracted from the literature. The models were then converted into one simulation package using C-sharp as the source code language by Elizabeth Post, Senior Computer Scientist at Lincoln Ventures Ltd, Lincoln, New Zealand and the author. This model was then used to investigate the nitrogen dynamics of a dairy farm: the relationship with pasture composition and whether small changes in pasture residuals make a difference to pasture composition and nitrogen dynamics. Two different simulations were run based on the management practice of Lincoln University Dairy farm (LUDF) over two dairy seasons (2002-03 and 2003-04) and validated against the data recorded on this farm. In 2002-03, 50 cows were over wintered and 580 cows were subsequently milked on 200ha. Post grazing residuals where maintained at 1600-1750KgDM/ha. In 2003-04, 125 cows were overwintered and 635 cows were milked on 200ha with post grazing residuals maintained at 1400KgDm/ha. All models operate on a daily time step. Within the pasture model composition is described by 9 state variables describing different components of the pasture and pasture growth is modelled mechanistically from a calculation of component photosynthesis. A further 9 state variables describe the nitrogen composition of the pasture components. The soil model is a variable two layer, mechanistic representation, parametised for the shallow, stony soils of LUDF. Soil water status is an input for the pasture model while water uptake by the growing plants affects the soil water balance within the soil model. Animal intake and production are modelled mechanistically with model cows described in terms of their age, genetic merit, body weight, breed, pregnancy status, conception date and body condition score. Each cow type produces a different quantity of urinary and faecal excretion which varies with dry matter intake, milk yield and the sodium and potassium status of the pasture. Excretal nitrogen composition is predicted within a separate model which calculates daily nitrogen excretion in faeces, urine and milk. Excretions are deposited randomly over the grazed area and account is taken of overlapping excretions that are created on the same day and overlaps that occur with older excretal patches deposited in previous grazing rounds. Each excretal patch has its own associated pasture, water and soil model reflecting the differences in nitrogen status between patches. Grazing preference is expressed within the model between different classes of excretal patch and between excretal patches and the base pasture and between clover and grass. Supplementary silage is conserved and fed according to the management schedule of LUDF. Cows calve, become pregnant and are dried off within the model according to the relevant records from LUDF. Cows are deemed to arrive on the farm on the day of calving and to leave on the day that drying off is finished (a 5 day procedure within the model), except for those cows that are overwintering which remain on the farm. The soil model has multiple nitrogen/carbon pools and is dynamically linked to all the other models. External nitrogen losses from the system are modelled as volatilisation, leaching and denitrification, with pasture nitrogen uptake from the soil model and fixation by clover from the atmosphere. Both the individual component models and the final assembled composite model were successful in matching the available data in terms of pasture and animal production, pasture composition, soil water balance and nitrogen status and external losses. The model indicates that the low residual, high stocking rate farm returns more excreta to the soil. However, this is countered by a reduction in the amount of dead material returned to the paddock and this reduces the relative size of the pool of nitrogen in the dead organic matter. This produces a relative lack of substrate for the soil microbes which are thus unable to exploit all of the nitrogen in the available pool. Soil ammonium and nitrate pools are also increased from the increase in faecal and urinary return so precipitating an immobilising flux from these larger pools to the smaller pool of nitrogen available to the soil microbes. However, the relative inability of the soil bacteria to fully exploit this means that the production of soil organic live matter and the resulting mineralising flux from the dead organic matter pool through the available pool to the ammonium and nitrate pools is reduced. The larger ammonium and nitrate pools will also be associated with increased external losses from the system as denitrification, leaching and volatilisation are increased. The increase in the clover percentage within the sward in 2003-04 led to greater nitrogen fixation and the model suggests that some of the extra nitrogen is effectively captured by the animals in increased production. However, the reduction in the return of dead matter coupled with an increase in excretal return and the consequent increase in the mineral nitrogen pools within the soil lead to greater losses of nitrogen from the soil.