Solving the conflict between intensive and extensive approaches: transect based sampling design for comparative studies on fine scale plant community organization

Background and Aims: Non-equilibrium ecological paradigm considers plant community as a complex dissipative system, which calls for a methodology with explicit representation of spatiotemporal patterns. However, recording vegetation patterns at this fine scale is time consuming and labour intensive. In contrast, understanding general rules of community organization and vegetation structure would require large number of comparative case studies. There is a clear trade-off between these intensive and extensive aspects in ecological applications. Here, we explore how field sampling techniques can be optimized compromizing between high resolution and large extent data collections. The coordinated distributed experiments and surveys based on these optimized sampling techniques might open new perspectives in comparative community ecology and macroecology. Materiai & Methods We used simulated data and field patterns recorded in form of spatial coordinates of plant individuals or presence of species in high resolution grids. Applying computerized resampling techniques we tested how coenostate variables will change by changing the sampling parameters (resolution, extent and the shape of sampled area). We used information theory models for analyses which represent complex community patterns (beta diversity of species combinations and species associations) as a function of spatial resolution. (Campetella et al. 2004). Main Results & Interpretations Results did not differ between high resolution grid data and spatial coordinate data. The absolute values of diversity and spatial dependence were similar between grids and transects, while the related characteristic scales slightly changed. Although scales were slightly biased when measured by transects, all ordering relations (i.e. differences between the compared vegetation types) remained invariant. Decreasing the spatial extent of samples resulted in strong increase of stochastic variance and produced artefacts. These problems were less pronounced when transects were used or the shape of grids become elongated. Comparing the effects of different sampling parameters, sample extent was the most critical. Using the same extent, transects give more representative data. Transect sampling was also much faster than other sampling methods. We concluded that resolution and extent could be optimized if long (50 m) transects of contagious 5 cm x 5 cm sampling units were used. This protocol was tested and proved to be applicable in a wide range of vegetation types including forest herb layer communities, grasslands in old fields, tall- and shortgrass steppes, mountain grasslands and semi-desert communities (Gosz et al. 2000, Virágh et al. 2008). We propose using this sampling design in future coordinated distributed experiments and surveys for studying non-equilibrium dynamics and assembly rules of vegetation in a more operative way and improving the predictability of vegetation processes.