Sampling grain influences trends in vegetation composition and diversity with time since fire in Australian heathland

Scale‐dependency of pattern and process is well‐understood for many ecological communities; however, the influence of spatial scale (sampling grain) in detecting temporal change in communities is less well‐understood. The temperate lowland heathlands of south‐east Australia are one of the most fire‐prone ecosystems on earth. Despite the extensive literature documenting the effect of time since fire on heathlands, we know little about how sampling grain influences trends in vegetation variables over time, and whether these trends are scale‐dependent. Using 3500 ha of heathland in the Gippsland Lakes Coastal Park, south‐east Australia, we investigated how above‐ground species composition and diversity, and trends in these variables with increasing time since fire, were influenced by sampling grain (1 m², 10 m², 100 m², 900 m², 1 ha, 4 ha). Sampling grain influenced patterns detected in vegetation variables and in some instances, significantly affected their relationship with time since fire. Richness decreased with time since fire, with mean richness decreasing at three of the four grains, while total richness decreased at half of the sampled grains. Evenness (J) decreased with increasing time since fire for all grains except 1 m². The decline in diversity (H) with time since fire appeared to be independent of scale, as all grains decreased significantly with increasing time since fire. Community heterogeneity demonstrated a weak response to time since fire across most grains. Changes in composition among young (0–6 years since fire), intermediate (9–19 years) and old (23–27 years) sites were dependent on sampling grain, with all grains exhibiting significant differences in composition, apart from the 1 m² grain and the 100 m² grain (presence/absence data). Overall, species composition, richness, evenness, diversity and community heterogeneity were dependent on the scale at which the vegetation was sampled. In addition, trends in many of these vegetation variables with increasing time since fire were scale‐dependent. This work provides strong evidence that sampling at multiple grains contributes substantially to understanding pattern and process in heathlands.