Sample sizes for estimating key ecosystem characteristics in a tropical terra firme rainforest

Abstract

This study evaluated the sample sizes necessary to estimate several soil and vegetation characteristics within 10% confidence intervals with 95% probability in three terra firme tropical rainforest sites. Across all three plots, the most spatially heterogeneous variables were measurements of total standing crop root mass, ground surface litter mass, litter fall, root growth and soil respiration which required, on average, 152, 105, 52, 45 and 28 samples, respectively to estimate mean values within 10% confidence intervals with 95% probability. Leaf area index measurements integrated canopy characteristics over a relatively large spatial area and therefore only required five samples, on average, to achieve the same degree of precision. Measurements of soil temperature, moisture, carbon and nitrogen content in the surface 30 cm soil layer displayed the lowest degree of spatial variation: requiring a maximum of seven samples to estimate mean values within 10% confidence intervals with 95% probability. This study, together with a review of data from similar ecosystems, suggests that standing crop root mass, root growth, litter fall and ground surface litter mass are usually acutely under-sampled, which could impede detection and interpretation of patterns and processes in these potentially important ecosystem characteristics. This information may assist researchers to design effective sampling strategies for field experiments, particularly in tropical forests. Show more