Theme 5: Drought--Oral Sessions

Description

Dryland ecosystems cover a large share of the world’s terrestrial surface. Deficiency and spatio-temporal variability of precipitation as well as low vegetation growth rates make dry rangelands prone to degradation, especially under changing climate and intensified land use. Degradation often occurs gradually but sometimes, a sudden and surprising shift from a healthy to a degraded rangeland can be observed, where perennial grasses are lost, and bare soil is exposed. If such changes are sudden and irreversible, they are coined a tipping point. Due to their abrupt appearance, it is a great challenge to discover early warning signals that precede the regime shifts. Theory predicts that variance and autocorrelation in state conditions could be used as early warning signals. However, these theoretical assumptions have rarely been tested in real ecosystems. Here, we use a data-based approach to contribute to filling this research gap using desertification processes in a semi-arid rangeland as a case study. In order to test the applicability of theoretical early warning signals for tipping points, we looked at a dataset from Widou, Senegal, that includes annual observations of rainfall, grazing intensity and primary production from 1981 – 2007. We analysed productivity-based metrics, such as rain use efficiency, in order to detect patterns that may precede a shift between alternate stable states. Strong signals of a regime shift were detected that were expressed in a sudden alteration of species composition and general decline of productivity after a drought. However, we did not find any changes in the theoretically proposed parameters that may reflect early warning signals for a critical transition, i.e. the regime shift was essentially unpredictable. We suggest that while the theory around tipping points and early recognition thereof may be robust, the applicability of theoretical concepts to the real world may be challenging.

Share

COinS
 

Towards Early Warning Signals for Desertification

Dryland ecosystems cover a large share of the world’s terrestrial surface. Deficiency and spatio-temporal variability of precipitation as well as low vegetation growth rates make dry rangelands prone to degradation, especially under changing climate and intensified land use. Degradation often occurs gradually but sometimes, a sudden and surprising shift from a healthy to a degraded rangeland can be observed, where perennial grasses are lost, and bare soil is exposed. If such changes are sudden and irreversible, they are coined a tipping point. Due to their abrupt appearance, it is a great challenge to discover early warning signals that precede the regime shifts. Theory predicts that variance and autocorrelation in state conditions could be used as early warning signals. However, these theoretical assumptions have rarely been tested in real ecosystems. Here, we use a data-based approach to contribute to filling this research gap using desertification processes in a semi-arid rangeland as a case study. In order to test the applicability of theoretical early warning signals for tipping points, we looked at a dataset from Widou, Senegal, that includes annual observations of rainfall, grazing intensity and primary production from 1981 – 2007. We analysed productivity-based metrics, such as rain use efficiency, in order to detect patterns that may precede a shift between alternate stable states. Strong signals of a regime shift were detected that were expressed in a sudden alteration of species composition and general decline of productivity after a drought. However, we did not find any changes in the theoretically proposed parameters that may reflect early warning signals for a critical transition, i.e. the regime shift was essentially unpredictable. We suggest that while the theory around tipping points and early recognition thereof may be robust, the applicability of theoretical concepts to the real world may be challenging.