An ED-based Protocol for Optimal Sampling of Biodiversity

Abstract

While conservation planning requires good biodiversity data, our knowledge of most living groups is scarce and patchy even in well-sampled regions. Therefore, we need methodologies for rapid assessments for particular groups and regions. Maps of any biodiversity surrogate can be interpolated from even a few well-known sites, but such places are usually lacking. We therefore propose a protocol for designing field surveys to obtain good coverage of pattern variations of biodiversity in a given region. To represent biodiversity patterns comprehensively, we use a rule step site-allocation procedure, partially based on Faith and Walker's ED criterion that takes environmental and spatial variation into account, together with other criteria such as survey costs. A preliminary assessment of the adequacy of this site sampling strategy is made. Then a set of complementary sites is selected for further sampling. Using the ED criterion, during the stepwise process a p-median analysis is applied both to an environmental distance matrix and to a spatial distance matrix, to maximize the amount of variation covered by our survey planning. This rule-set allocation procedure is integrated into a continuous sampling design protocol directed to ensure we can sample all biodiversity of a region. This protocol requires the gathering of both biological and environmental information, an assessment of previously available information, the choice of sampling methods and dates, and a continuous assessment of the success of the survey being carried out. An example of the application of this protocol to the survey design of dung beetle (Coleoptera, Scarabaeoidea) diversity in the Comunidad de Madrid (Spain) is included.

Increased Diversity of Arbuscular Mycorrhizal Fungi in a Long-Term Field Experiment via Application of Organic Amendments to a Semiarid Degraded Soil

“…3) showed that the number of analyzed root samples was sufficient to detect the majority of fungal types present in the roots as the curve approaches saturation. The estimated-richness curve showed the same tendency as the observed-richness curve, and its slope had a value of 0.018, which was lower than 0.05 (slopes lower than a cutoff value such as 0.05 are complete enough to consider a taxonomic unit or phylotype to be well sampled for studies of biodiversity) (23).…”

Distribution and Endemism of Angiosperms in the Atlantic Forest

“…The Jackknife 1 estimator has been shown to give better results than the other incidence-based estimators, since both seem to be less affected by grain size (Hortal et al 2006). In every grid cell, the number of database records was used as a surrogate of sampling effort (Hortal & Lobo 2005). We use a biodiversity-oriented Geographical information system (DIVA-GIS 7.4, http://www.diva-gis.org/) for the calculation of Jackknife1 estimator of species numbers per grid cell.…”

Can Species Richness Patterns Be Interpolated From a Limited Number of Well-Known Areas? Mapping Diversity Using GLM and Kriging

“…We merged the data from three of the 255 grid cells used by them, which had less than 15% of land surface, with their adjacent UTM50. Lobo & Martín-Piera (2002) used species accumulation curves to relate the sampling effort carried out on each cell and the number of species discovered (see also Hortal & Lobo 2005), identifying 82 UTM50 as being well-sampled (Lobo & Martín-Piera 2002). They later excluded seven of these cells, because they either pertained to the Balearic Islands or were identified as outliers due to oversampling (around Madrid and Barcelona).…”

“…We assessed the coverage of the environmental and geographical variability of the Iberian Peninsula provided by the well-sampled UTM50 by means of their ED coverage (that is, coverage of the overall Environmental Diversity; see Hortal & Lobo 2005 and references therein). Here, the overall coverage of the environmental and/or spatial variability of a territory provided by a selected subset of areas is calculated as the sum of all the distances from each non-selected area to the selected subset; the lower such sum of distances, the larger is the coverage of the regional variability, in a typically decreasing curve.…”