Estimating tree species diversity across geographic scales

The relationship between number of species and area observed has been described using numerous approaches and has been discussed for more than a century. The general objectives of our study were fourfold: (1) to evaluate the behaviour of species–area curves across geographic scales, (2) to determine sample sizes necessary to produce acceptably precise estimates of tree species diversity, (3) to evaluate relationships among estimates of tree species diversity for local to large geographic scales, and (4) to determine whether the proportion of native tree species may be precisely estimated using sample sizes smaller than those necessary to estimate total tree species diversity. Such investigations are necessary to improve biodiversity monitoring at large scales. The analytical approaches included Monte Carlo bootstrap simulations and two geospatial methods and relied on a database populated with data for more than 12,000 national forest inventory plots (NFI) from 16 regional units, 13 European countries and three ecoprovinces of the United States of America (USA). Four primary results were obtained. First, tree species diversity may be precisely estimated using observations for a random subsample of 2,000–4,000 NFI plots. Second, large sample sizes are necessary to estimate tree species diversity for regional units or forest categories, except possibly for boreal forests for which the number of tree species is small. Third, estimates of the proportion of native tree species may be precisely estimated using tree species information for a random sample of approximately 30 NFI plots. Finally, our estimated species–area curves show that the curve shapes and relationships among estimates of tree species diversity at large scales clearly depend on the relative geographic location of the anchor regional unit (a European country or ecoprovince of the USA) relative to the other regional units. None of the well-known models for species–area curves adequately describes our results. The conclusion was that the uncertainties of the estimates reflect the unfavourable state of global biodiversity monitoring of species groups. The total numbers of tree species in our regional units, which cannot be adequately estimated, are small relative to other tree species-rich regions. Consequently, monitoring of tree species diversity is currently a highly uncertain enterprise at large scales.