Abstract
In 2008, I had the chance to gather my first experiences on the African continent during an internship in Senegal. Returning home with a passion for the African nature, the landscapes and especially the people, the decision to work in this region again in the course of my further studies matured in my mind. This study, conducted as a diploma thesis at the University of Greifswald, investigates plant diversity patterns in the Namibian savanna across multiple spatial scales.
Introduction
Ecologists face a fundamental "scaling" problem: it is impossible to count every plant in a large landscape, necessitating the use of small samples to estimate total biodiversity. However, different mathematical models—specifically Species-Area Relationships (SARs) and Species-Sampling Relationships (SSRs)—often yield divergent results.
Semi-arid savannas are particularly sensitive to land use. In central Namibia, "fence-line contrasts" provide a natural experiment where adjacent farms experience vastly different grazing intensities. This study addresses how grazing pressure affects plant diversity across scales ranging from 0.01 m² to 1 km² and evaluates which statistical methods provide the most accurate regional biodiversity estimates.
Methodology
The research employed a multi-scale nested plot design on two adjacent farms: Narais (lightly grazed) and Duruchaus (heavily grazed).
Fieldwork and Data Collection
Forty main plots of 1,000 m² each were sampled, containing nested subplots down to 0.01 m². Primary vegetation data was collected in March 2009, totaling 20,000 m² of sampled area. This was supplemented by long-term monitoring data (2005–2009) from the BIOTA Southern Africa project to validate extrapolation models.
Modeling and Evaluation
The study fitted 12 different SAR models and 6 SSR models to the data. Model performance was evaluated using:
- Akaike Information Criterion (AICc): To assess goodness-of-fit.
- Logarithmic Error of Extrapolation (LEE): To test the accuracy of richness predictions beyond the sampled area.
- Time-Effort Analysis: Comparing standard 15–30 minute monitoring protocols against intensive ~2-hour surveys to quantify overlooked diversity.
Results and Discussion
The Impact of Grazing
Contrary to expectations, heavy grazing did not significantly reduce total species richness. Instead, it induced a shift in community composition. The heavily grazed site (Duruchaus) was dominated by grazing-tolerant and annual species. Interestingly, at the smallest scale (0.1 m²), the heavily grazed site showed higher average richness (5.67 species) compared to the lightly grazed site (4.48 species), likely due to the intermediate disturbance hypothesis where annuals fill gaps created by grazing. Conversely, the lightly grazed site (Narais) exhibited higher "evenness," indicating a more balanced species distribution.
Scaling and Model Performance
The study found that area-based models (SARs) were significantly superior to sample-based models (SSRs) or non-parametric estimators (e.g., Chao 2) for upscaling local observations to a 1 km² area. Non-parametric estimators consistently underestimated true diversity. Among the models tested, the classic Arrhenius power function ($S = cA^z$) emerged as the most reliable tool for both goodness-of-fit and extrapolation capability.
Sampling Completeness
A total of 213 plant species were observed. Models estimated the true richness of the 1 km² observatories to be between 260 and 350 species. Standard 30-minute surveys captured only approximately 63% of the estimated total species; reaching 85% completeness required roughly two hours of intensive searching per hectare.
Conclusion and Future Directions
While the power function remains a robust tool for biodiversity estimation, this study highlights the limitations of single-year snapshots in highly variable savanna ecosystems. The random sampling scheme may miss rare "azonal" habitats like pans and drainage lines. Future monitoring should prioritize cumulative data collection over multiple years to account for the ephemeral nature of savanna annuals and the influence of rainfall variability.
References
Peters, J. (2010). Plant diversity patterns at different spatial scales in a semi-arid savanna ecosystem in central Namibia. Diploma Thesis, Ernst-Moritz-Arndt-Universität Greifswald.
Summary updated