Plant inventories are at the heart of conservation efforts. Despite their obvious conservation values, properties of these datasets are poorly understood. We use plant databases from three different well-established inventories [rare plants in California (CA), Spanish threatened plants (SP) and the Proteaceae in South Africa (SA)] to explore the behavior of large data sets in facilitating the link between current field surveys and future standardized monitoring methods. We analyze area frequency curves of the species area size for each data set and for a series of extracted databases from each inventory. Our results show that all field surveys produced left-skewed frequency distributions. A lognormal distribution is better fitted by SA, followed by CA and finally by SP, which is least suited to a lognormal fit. Using the most threatened portion of the three floras, these general patterns still apply. Secondly, a minimum sample analysis indicates that precision increases according to sample size. Proportionally, CA data require less sampling effort than the Spanish pool and the latter require less than do SA in order to get a clear monitoring trend. Larger skewness values are related to inventories with wider scope. SA Proteas display the most skewed distribution. Skewness in California may be explained not only by the nature and scope of the inventory but also by the scale used for mapping. The Spanish database is also affected by surveyor bias towards the most endangered portion of the data set. Monitoring should take into account the original nature of each inventory. Particular inventory methods and scope may produce different outputs, constraining future monitoring programs. Key aspects are skewness and variation, and both combined could identify inventories in need of better data collection practices for more precise estimates of changes in biodiversity.
Read full abstract