Search

Long-term agroecosystem experiments can be defined as large-scale field experiments more than 20 years old that study crop production, nutrient cycling, and environmental impacts of agriculture. They provide a resource for evaluating biological, biogeochemical, and environmental dimensions of agricultural sustainability; for predicting future global changes; and for validating model competence and performance. A systematic assessment is needed to determine the merits of all known experiments and to identify any that may exist in tropical and subtropical environments. The establishment of an international network to coordinate data collection and link sites would facilitate more precise prediction of agroecosystem sustainability and future global change.

This paper describes Century Soil Organic Matter (SOM) Model simulations of seven long-term data sets that are the subject of this special issue. We found that Century successfully simulates SOM C across a variety of land use and climate types. Simulations of SOM were most successful in grass and crop systems. This exercise highlights a structural limitation of Century in simulating SOM in a forest with a developed litter layer. Simulations of tree biomass distributions, however, were generally successful. The model failed to capture extreme values of yield and N offtake, although annual averages were quite similar between observations and simulations, leading to reasonable estimates of SOM C. Average yields and SOM were generally higher in amended treatments or rotations including an N-fixing component such as alfalfa. The model successfully predicted SOM dynamics across climates, land use types, and treatments. This suggests that Century is a useful tool for ecosystem studies, particularly those focused on SOM dynamics.

Simulations of long-term (> 20 year) soil organic carbon (SOC) dynamics by the DNDC model were compared with field observations at 11 plots in 5 field stations in Europe and Australia. The exercise was part of a NATO-sponsored workshop on long-term monitoring and modeling of soil organic matter. Eight of the eleven plots were cultivated cropland and three were grassland (harvested for hay). There were a range of fertilizer and manure treatments, as well as crop rotation sequences. Significant loss in SOC was observed at two plots in Australia where a grassland had been converted to cultivated cropland in 1925. Both field data and model simulations showed the plots reaching a new SOC equilibrium at about 44% of the 1925 levels. Equilibrium levels depended on crop rotation sequence, with higher SOC for the plot with less frequent fallowing. At one permanent grassland site at Rothamsted, UK, a large decline and recovery in SOC was observed in the field, but not in the model simulation. For all other cases, both field and model data showed relatively small changes in SOC, though field data tended to be more variable, perhaps due to variability in both crop and weed yield, and in residue management. Mean percent differences between simulated and measured SOC were 0.07% or less (as percent by weight, kilogram SOC/kilogram soil) for all but one of the plots simulated.

As part of a model evaluation exercise, RothC-26.3, a model for the turnover of organic carbon in non-waterlogged soils, was fitted to measurements of organic carbon from 18 different experimental treatments on 6 long-term experimental sites in Germany, England, the USA, the Czech Republic and Australia. In the fitting process, the model was first run with an annual return of plant C that had been selected iteratively to give the carbon content of the soil at the start of each experiment. This was done for the soil and climate of each site. If the radiocarbon content of the soil organic matter was known, the inert organic carbon (IOM) content of the soil was also calculated for the start of the experiment. Using these carbon and radiocarbon contents as a starting point, the model was then run for each of the experimental treatments to be fitted, using iteratively selected values for the annual return of plant materials to the soil. The value used for each treatment was selected to optimise the fit between modelled and measured data over the whole experimental period: fitting was done by eye. Thus fitted, RothC-26.3 gave an acceptable approximation to the measurements for 14 of the treatments, bearing in mind the experimental errors in measuring soil organic carbon on a per hectare basis. With four of the treatments (Highfield Bare Fallow, Park Grass plot 13d, Ruzyně farmyard manure plot and Tamworth rotation 5), the fit was less satisfactory.

CANDY (CArbon and Nitrogen DYnamics) is a simulation system based on long-term experiments of organic matter turnover and nitrogen dynamics at Bad Lauchstädt, Germany. Key driving variables are soil physical properties, meteorological data and management information. The main application of the CANDY model is the calculation of short-term dynamics of nitrogen transformation and long-term dynamics of organic matter turnover in arable soils. This paper concentrates on the evaluation of the model in simulating carbon dynamics in long-term experiments representing different land uses and very different geographical sites. The experimental data came from data sets that were made available to modellers at a workshop held at Rothamsted in 1995. In this paper we describe how the different data sets were modelled and provide a qualitative assessment of model performance. The performance of several models, including CANDY, are compared quantitatively in Smith et al. (1997). Our results show that the mathematical basis of the model, its consideration of a biological time base and its calculation of the `reproducing carbon' are applicable over a wide range of sites and land-use scenarios. Most of the standard parameters can be used for other sites and land-use systems.

DAISY, a mechanistic model for the simulation of crop production, soil water and N dynamics in agro-ecosystems, was used for simulating trends in total soil carbon (C) for various long-term experiments as part of a model evaluation exercise. Of the seven core datasets in the global SOMNET database, DAISY could be applied to the four arable sites and, with certain critical assumptions, to the single grassland site. The model was, however, not applicable to the two forest sites, because currently DAISY cannot simulate soil water and temperature dynamics in a forest system and does not incorporate leaching of dissolved organic substances, faunal litter incorporation or the effect of low pH on decomposition. The DAISY crop submodel produced some overestimation of crop yields, especially at low N input levels, and simulated root C input to the soil may thus represent a slight overestimation. This may have been balanced, however, by the fact that no account is taken of rhizodeposition or root turnover in the current version of the DAISY model. For this reason too, plant-derived C input in perennial crops (pastures) had to be estimated and input separately. Under these critical assumptions, the model was capable of simulating long-term trends in total soil C quite well for all included sites and treatments (overall correlation, R2 = 0.88), except for the Bad Lauchstädt arable site. It was evident that sufficient experimental data on belowground, plant-derived C inputs was not available, especially with perennial crops. Future research in this area should thus have high priority, because such data are a prerequisite if soil organic matter models are to incorporate and improve the simulation of this very important mechanism of soil C input. The partitioning and parameterisation of C between pools in the DAISY model has been discussed and indications were found that the parameterisation of the microbial biomass pools in the model are somewhat inadequate.

The Permanent Rotation Trial at the Waite Agricultural Research Institute in South Australia was established on a red-brown earth in 1925, with predominately cereal-long fallow rotations on 34 adjacent plots. The trial was upgraded in 1948 to include a greater proportion of pasture leys in the rotations and currently contains 11 treatments. The trial is unreplicated; however, each phase of a sequence is represented each year. Seven of the original rotations have remained in an unbroken sequence since 1925: continuous wheat (W), wheat-fallow (WF), wheat-peas (WPe), wheat-pasture-fallow (WPaF), wheatoats- fallow (WOF), wheat-barley-peas (WBPe), wheat-oats-pasture-fallow (WOPaF). For the 11 rotations, soil organic carbon (SOC) in the top 10 cm declined from 2.75% in 1925 to a mean value of 1.56% in 1993. One plot, which had reverted to permanent pasture in 1950, showed the smallest decline with an SOC content of 2.46% in 1993. The greatest declines in SOC were in the 4 original rotations that included fallow phases in the sequence (mean value of 1.22%). In the WF rotation the SOC content had declined from 2.75 to 1.04% during 68 years of cropping. Associated yield decreases showed that the treatment could not sustain production. Soil organic C declined linearly with increasing frequency of fallows and decreasing frequency of pasture in the rotations. Average grain yields (1925-93) in the 7 original sequences ranged from 2.64 t/ha in WOPaF to 0.89 t/ha in the continuous W plot. The linear decline in yields for WBPe, WPaF, WPe, and WOF treatments indicate a convergence in the 1990s under current management, with an average yield of 1.54 t/ha in 1993 and average SOC in the top 10 cm of 1.32%. We hypothesise that the gradual increase in grain yields from the continuous W plot since the 1960s is the result of a gradual build-up of light fraction organic material, which assists in the maintenence of structure and nutrient availability.