Untitled

Sampling intraspecific variability in leaf functional traits: Practical suggestions to maximize collected information

“…A very actual debate in "trait-based ecology" is focused on the importance and relative magnitude of interspecific (BTV, B stands for between species) and intraspecific variability (ITV) (Albert et al, 2010). While sources and effects of BTV on functional traitbased studies have been widely investigated (Diaz et al, 2004;Wright et al, 2005), the contribution of ITV to the total variability of a trait has been underestimated (Violle et al, 2012). ITV is defined as the overall variability of trait values and trait syndromes (set of trait values including trait trade-offs) expressed by individuals within a species (Albert, Grassein, Schurr, Vieilledent, & Violle, 2011).…”

Vegetation mapping from high-resolution satellite images in the heterogeneous arid environments of Socotra Island (Yemen)

“…Each subclass is defined by a multivariate Gaussian distribution, and at the same time the combined mixture of distributions adapts to nonlinear patterns in the data; thus, the Gaussian mixture for each class does not resemble a Gaussian distribution. 5,34 We estimated the number of subclasses directly from the data, with the advantage of an objective determination of the tightness of fit that prevents over-and underfitting. These operations were executed in GRASS through the module i.gensigset.…”

“…3 Several studies have proven that traditional methods (e.g., field surveys, literature reviews, map interpretation, and ancillary data analysis) are not effective to map vegetation cover, since they are time-consuming, date-lagged, and often too expensive. [4][5][6] Conversely, remote sensing (RS) represents a practical and economical instrument to study vegetation cover 4,7 and has been applied to map vegetation cover from local to global scales over the last three decades. 6 RS has been widely applied in vegetation and land cover mapping of arid environments, especially when combined with thorough ground-truthing.…”

“…6,11,15 In these cases, traditional classification approaches like maximum likelihood (ML) and single-class signatures proved to be equally ineffective given the impossibility to describe heterogeneous classes and the high level of map fragmentation and misclassifications obtained in the case of mosaicked environments. 5,16,17 Therefore, ecologically meaningful vegetation maps need to be based on thorough vegetation surveys/analysis, hyperspatial/hyperspectral satellite data, and precise and flexible classification approaches in order to provide a reliable reference for monitoring, management, and planning. 6,15,17,18 Kràl and Pavlis 9 first attempted to produce a land cover map of Socotra Island (Yemen), a global biodiversity hotspot located in the Indian Ocean, through an ML classification of LANDSAT coarse-resolution images based on field verifications.…”

“…19,20 In the present study, in order to produce an ecologically accurate vegetation map of Socotra, a high-resolution, multispectral satellite image was classified based on a sequential segmentation process particularly suited to complex vegetation mosaics. 5,11,12 This classification was compared to the traditional ML approach by assessing the percentage of accuracy reached on the basis of a reference dataset obtained from a statistical analysis of more than 300 vegetation field surveys.…”