Desertification of high latitude ecosystems: conceptual models, time-series analyses and experiments

Ecosystem degradation in Iceland has been severe since man arrived 1100 years ago. Birch woodlands cover has declined from 25% of the land area, to only 1%. The deforestation is considered to be the initial stage in the land degradation process, followed by surface destabilization, and later erosion. The objective of this study was to quantify and evaluate factors that contribute to the early stages of land degradation in Icelandic ecosystems. Specific objectives were to improve our understanding of how livestock grazing might initiate early degradation stages, elucidate field-based landscape metrics useful for characterizing degradation stages, and to determine if landscape metrics obtained from remote sensing data can be used to detect landscape structure changes and identify degraded and at risk rangelands in real time over extensive and remote areas. A State-and-Transition conceptual model was constructed for the experimental area to identify potential key processes in the degradation sequence, and to formalize research questions. Experimental plots were established in five plant community types representing a space-for-time degradation sequence. Birch seedling (Betula pubescens Ehrh.) growth and survival was reduced with repeated clipping treatment applied to simulate browsing, but the amount of decline varied with plant community type. This suggests that continuous grazing may contribute to deforestation, as regeneration will be reduced over time. Intense grazing treatments, simulating both grazing and trampling, increased surface instability and soil loss compared to grazing only or control, suggesting that intense grazing may contribute to surface destabilization and therefore to land degradation. Erosion appeared to be active in the most intense treatments, also within the woodlands. The data indicate that the woodlands may have lower resilience than the other plant communities as treatment effects appeared quicker there. The woodlands may thus be particularly vulnerable to intense grazing. The landscape metrics used to quantify changes in landscape surface properties over a 51 year period yielded inconclusive results, either because of data limitations or because of non-detectable erosion activity. The results do generally support the proposed S&T model for the experimental area. It is concluded that grazing may contribute to woodland decline, and intensify degradation processes.