Modelling climate change impacts on European grassland-based livestock systems

Abstract

Climate change is leading to higher temperatures and altered rainfall patterns across Europe. These changes are likely to have major impacts on plant life. This is particularly relevant for livestock production systems which are dependent on grass and forage. Farmers need to know what they can expect in the future so that they can be well prepared and ensure that their livestock will have enough to eat. This thesis aims to quantify the impacts of rising atmospheric CO2 concentrations, higher temperatures and changes in water availability on the yield and protein content of European grasslands.

The first approach used was a meta-analysis. Data from experiments in which the climate had been artificially altered was collected and divided according to geographic region (Alpine, Atlantic, continental, northern and southern) and plant type (graminoids, legumes, forbs and shrubs). Using Markov Chain Monte Carlo (MCMC) simulations, mixed models were developed to estimate the expected changes to plant yield and protein (i.e. nitrogen (N)) concentration under different climatic changes. The results showed that areas predicted to become warmer and wetter (i.e. northern Europe and parts of Alpine and continental Europe) will benefit from higher plant yields, but reduced plant N concentration. Areas which will become warmer and drier (i.e. southern Europe and parts of continental Europe) will see decreases in both yield and N concentration. The Atlantic region is the area where climate change is expected to be the least extreme and the effects on plant life will be relatively minor. Shrubs will particularly benefit from rising atmospheric CO2 concentrations, though will also suffer large decreases in N concentration, as will forbs.

The next approach considered different methodologies for modelling grassland yield and N yield. One method involved developing a statistical model using data from long-term grassland experiments across Europe. Through stepwise linear regression, equations were developed to model grassland yield and N yield based on various weather and managerial variables. The other method used a pre-existing process-based model (Century), which was applied to six sites across Europe. Both approaches produced reasonable estimates of grassland yield and N yield. The prediction error was lower for the Century model while the regression methodology produced better correlations between observations and predictions. Both models were quite sensitive to uncertainties in weather parameters, particularly precipitation, with little sensitivity to soil properties. Overall, the regression approach was found to be suitable for considering general trends over large spatial scales, while the Century model was more appropriate for local-scale analysis.

The two models described above were used to quantify the effects of two different climate change scenarios (one midrange and one more extreme) on the five European regions listed above. The two models generally produced similar predictions, indicating that grassland yields will increase in most areas though there may be slight decreases in southern Europe. Also, plant N concentrations will decrease. Generally permanent grasslands responded more positively to climate change than temporary ones. The impact of climate change tends to be less than the impact of fertiliser, geographic region or grassland type, suggesting that appropriate changes to grassland management practices should be able to mitigate the negative effects of climate change.

The modelling described above was all performed using a monthly time-step. This is computationally efficient, but means that short-term extreme weather events are not accounted for. Extreme weather events such as heavy rainfall, droughts and heat waves are predicted to become both more frequent and more intense in the future and it is important to consider the impacts they will have on grasslands and therefore livestock.

Two methodologies were used to quantify the effects of extreme weather events on grasslands. The first uses multiple regression analysis and incorporates terms such as ‘number of days in a month with temperature greater than 30°’ to account for weather extremes. The equations developed had a good fit with observed data. They were found to be predominantly sensitive to uncertainties in precipitation rather than in temperature or grassland species composition. Two projected future weather datasets were applied to the equations; both followed the same climate change scenario, but one included extreme events and the other was smoothed to reduce the extremes. Comparing the model outputs from the two datasets showed that smoothing the data increased the predicted yields and N yields, demonstrating that extreme weather events are detrimental to grasslands. In general, the yield of temporary grasslands decreased over time, while for permanent grasslands it increased. There was little change in N yield over time.

The other methodology used the pre-existing process-based model DailyDayCent, which is very similar to the Century model, but is based on a daily rather than a monthly time-step. DailyDayCent was applied to six sites across Europe and was found to have reasonably good fit, though struggled to capture inter-annual variability. The model was predominantly sensitive to uncertainties in rainfall measurements rather than temperature. Two climate change datasets, with and without extreme events, were applied to the model for each of the six sites. Predicted yields and N yields were similar to those found with the Century model. The presence or absence of extreme events usually had little effect, but this may have been due to limitations of the model. The exception was for a site in southern Europe, where the presence of extreme events led to increases in yield and N yield in the short-term, but large decreases in the long-term.

Overall, grassland yields are expected to increase in the future in response to climate change (except possibly in southern Europe), particularly for permanent grasslands, while plant N concentration will decrease. Increased yields are generally good for livestock, though reduced N concentrations indicate that grazing animals will need to have a higher intake in order to receive the same amount of protein. Extreme weather events are an important consideration, leading to reductions in grassland yield and N yield. Farmers need to be prepared to meet the challenges presented by such events, for example through using more resilient plant species or increasing plant species richness.