Informatics software for the ecologist's toolbox: A basic example

The Use of Hyperspectral Visible and Near Infrared Reflectance Spectroscopy for the Characterization of Salt-Affected Soils in the Harran Plain, Turkey

“…The overall accuracies of 60% (j ¼ 0.74, p < 0.05) and 42% (j ¼ 0.16, p < 0.1) were obtained for training and testing, respectively. The lower accuracies obtained could be due to the low number of samples used in the training data set, and better results with CART are generally achieved when using more samples (Williams & Poff, 2006). Also, uneven …”

Prediction of stream fish assemblages from land use characteristics: implications for cost-effective design of monitoring programmes

“…The Kappa statistics were calculated from a contingency table with observed and expected predictions. A p value for a test of significance (Z ) for the hypothesis that Kappa = 0 (predictions not better than chance alone) was also calculated (Williams and Poff 2006). Thirdly, model accuracy was evaluated using Pearson correlations between the observed (classifications from classification trees) versus expected (original assemblage membership) member class (Usio 2007).…”

Beyond data management: how ecoinformatics can benefit environmental monitoring programs

“…The ecological condition of stream benthic macroinvertebrates has also been evaluated with self-organizing maps (Horrigan and Baird 2006;Song et al 2006), a data visualization technique that reduces the dimensions of data through the use of self-organizing neural networks. Such tools help ecologists without extensive knowledge of computational science to extract more information from multidimensional, interrelated datasets (Williams and Poff 2006). When observing a disturbed community, it is often difficult to identify what stressor(s) caused the condition; Van den Brink et al (2006) addressed this with a tool that links taxa sensitivity to stressors database with a second database on species traits such as water breathing ability, body size, and maximum age.…”

“…For instance, Pascoe et al (2006) developed BugML, an XML standard for sharing aquatic biomonitoring data from distributed data sources in order to construct a national indicator of ecological condition of freshwater ecosystems. Williams and Poff (2006) evaluated the application of artificial neural networks, evolutionary algorithms, and classification/regression trees to the USEPA's Environmental Monitoring and Assessment Program stream monitoring data in the U.S. mid-Atlantic area to calculate ecological indices. The ecological condition of stream benthic macroinvertebrates has also been evaluated with self-organizing maps (Horrigan and Baird 2006;Song et al 2006), a data visualization technique that reduces the dimensions of data through the use of self-organizing neural networks.…”