Get reliable rangeland science

We attempted to ascertain the most effective sample size for monitoring serial changes in the herbaceous layer of fragmented and neglected grasslands that are responding to annual veld burning. We conducted a sampling exercise in a coastal grassland invaded by the alien shrub Chromolaena odorata (chromolaena) and other indigenous woody species, from 1992 to 1998. We used permanent transects (samples) of differing length, ranging from 15?30m. Each transect was made up of contiguous 1m × 1m quadrats (sample units). In each quadrat, we recorded the species present and counted the number of woody alien plants. Chromolaena diminished under annual burning. Species richness and turnover increased in all transects over time. The 25m transect was as efficient as the 30m transect; however, the latter was influenced by an edge effect, confounding the determination of an optimum sample size for monitoring species richness. Species-accumulation curves were unable to predict alpha diversity at scales larger than transect sizes. Species composition was not affected by transect size; transects of 15m offered a 40?50% saving on sampling effort over 25m and 30m transects, respectively. Evaluations of sampling efficiency and precision are also discussed.