unlimited diffusion of knowledge, a simplification used to concentrate on the
coupled aspects of the model. In future simulations, an assessment of adoption
rates would need to be included. Still, even at this early stage, an important benefit
of the coupled model is its ability to depict farmers’ strategy changes from
extensification under decreasing yields to intensification under decreasing fertilizer
prices.
10.8.4.2 Effects of Soil Conservation
Our results show that in the coupled model farmers chose soil conservation to gain
higher yields, as long as this was profitable due to high fertilizer costs and when
competitive as compared to other crops. In contrast to the coupled model, with
continuously changing land cover, the standalone LUCIA model showed that in the
medium term soil conservation led to reduced erosion and higher yields.
As suggested by the much higher soil erosion rates in the standalone runs, soil
conservation measures did make a difference regarding erosion, but in the coupled
model these soil conserving effects were partially compensated for by the changing
land cover, or distorted by co-variation with fertilizer effects.
Although soil conservation was successfully applied on maize plots, both model
versions agreed that soil conservation did not have a major impact at the catchment
scale, owing to the relatively small area under maize (less than 5%) when compared
to other land cover types. In particular, areas under tree crops remained relatively
stable over time and did not contribute substantially to erosion. Only a small portion
of the large amounts of sediment simulated at the catchment outflow stemmed from
the respective cells; reflecting that the model does not distinguish between eroded
material native to a pixel and sediment passing through, having originated elsewhere in the watershed.
10.8.4.3 Spatial Aspects and Model Complexity
Land use change implemented in the coupled model system led to substantially
different outputs, e.g., reduced soil erosion and higher yields of unfertilized maize
after 5 years – when compared to the LUCIA standalone simulation, showing the
added value of coupling. Agents were able to react to changes in soil productivity,
at least at the farm level, and leave unproductive plots fallow, as a farmer would do
in reality. This may also have been the reason for lower spatial variability, while in
the standalone model maize was still grown even when the soil was no longer fertile
enough for sustained production.
In conclusion, the greater the number of variables, levels of complexity and thus
degrees of freedom there are, the more accurate the representation of reality will be
if the model system is well calibrated. As a trade-off, this requires more input data
and more effort in terms of interpreting the model outputs (Marohn et al. 2012).
While particularly non-linear cause-effect relationships can be represented to a
10 Integrated Modeling of Agricultural Systems in Mountainous Areas
425
coupled aspects of the model. In future simulations, an assessment of adoption
rates would need to be included. Still, even at this early stage, an important benefit
of the coupled model is its ability to depict farmers’ strategy changes from
extensification under decreasing yields to intensification under decreasing fertilizer
prices.
10.8.4.2 Effects of Soil Conservation
Our results show that in the coupled model farmers chose soil conservation to gain
higher yields, as long as this was profitable due to high fertilizer costs and when
competitive as compared to other crops. In contrast to the coupled model, with
continuously changing land cover, the standalone LUCIA model showed that in the
medium term soil conservation led to reduced erosion and higher yields.
As suggested by the much higher soil erosion rates in the standalone runs, soil
conservation measures did make a difference regarding erosion, but in the coupled
model these soil conserving effects were partially compensated for by the changing
land cover, or distorted by co-variation with fertilizer effects.
Although soil conservation was successfully applied on maize plots, both model
versions agreed that soil conservation did not have a major impact at the catchment
scale, owing to the relatively small area under maize (less than 5%) when compared
to other land cover types. In particular, areas under tree crops remained relatively
stable over time and did not contribute substantially to erosion. Only a small portion
of the large amounts of sediment simulated at the catchment outflow stemmed from
the respective cells; reflecting that the model does not distinguish between eroded
material native to a pixel and sediment passing through, having originated elsewhere in the watershed.
10.8.4.3 Spatial Aspects and Model Complexity
Land use change implemented in the coupled model system led to substantially
different outputs, e.g., reduced soil erosion and higher yields of unfertilized maize
after 5 years – when compared to the LUCIA standalone simulation, showing the
added value of coupling. Agents were able to react to changes in soil productivity,
at least at the farm level, and leave unproductive plots fallow, as a farmer would do
in reality. This may also have been the reason for lower spatial variability, while in
the standalone model maize was still grown even when the soil was no longer fertile
enough for sustained production.
In conclusion, the greater the number of variables, levels of complexity and thus
degrees of freedom there are, the more accurate the representation of reality will be
if the model system is well calibrated. As a trade-off, this requires more input data
and more effort in terms of interpreting the model outputs (Marohn et al. 2012).
While particularly non-linear cause-effect relationships can be represented to a
10 Integrated Modeling of Agricultural Systems in Mountainous Areas
425