Binary Discriminant Analysis Research Articles

Geochemical correlation of bentonites from Milos Island, Aegean, Greece

AbstractMajor and trace chemical elements from eleven Lower Pleistocene bentonite deposits from Eastern Milos Island, Aegean, Greece, were used for geochemical fingerprinting of the bentonites, by means of scatter binary and ternary plots and canonical discriminant analysis. The trace elements were selected according to their immobility during bentonite formation and subsequent hydrothermal alteration. The bentonites can be separated into at least four geographically distinct groups, each consisting of materials with similar geochemical affinities. The deposits of each group were derived from protoliths erupted from different volcanic centres, suggesting the existence of at least two distinct volcanic provinces, having rhyolitic and andesitic affinities respectively. Several deposits consist of more than one bentonite horizon, indicating that in some volcanic centres the character of volcanism changed with time. Geochemical fingerprinting, supplemented with geological, mineralogical, and textural evidence, may be useful for correlation and exploration of bentonites, in terrains with complex geological histories.

Response of high mountain landscape to topographic variables: Central pyrenees

An objective method for inductively modelling the distribution of mountain land units using GIS managed topographic variables is presented. The landscape of a small high mountain catchment in the Spanish Pyrenees, covered with grassland, was classified in ten land units by hierarchical agglomerative clustering, using a sample of 194 random plots, in which classes of vegetation, soils and landforms were defined. Additionally, seven layers of topographic variables (altitude, slope angle, aspect, solar radiation, topographic wetness index, specific catchment area, and regolith thickness) were created from a Digital Elevation Model. The affinity of each land unit to the topographic variables was calculated using Binary Discriminant Analysis (BDA), after dichotomising the latter around their mean values. Then, the distribution of each land unit was predicted by boolean operations combining step by step distributions for the seven topographic variables ordered, for each unit, after the absolute values of the Haberman’s residuals in BDA. The predicted distributions were tested (χ2) against that of the observed sampling plots. From the original ten land units, the distributions of eight of them were successfully predicted (four are related to the slope sequence, two reflect the water accumulation in the soil, and two respond to geomorphic processes) while the remaining two had to be rejected. Part of the catchment (39%) was not assigned to any land unit, probably because more distributed variables accounting for snow distribution are necessary.

Effect of fire intensity on understory composition and diversity in aKalmia-dominated oak forest, New England, USA

This study investigates the understory dynamics of two mixed-oak stands following fire of varying intensity. Composition and diversity of woody and herbaceous species in the understory were measured in two stands 7–8 years after a prescribed burn. On both sites, unburned areas, low-intensity fire areas, and areas where the overstory had been severely damaged were measured. Patterns of species presence and absence following fire were consistent with an initial-floristics model. Most species increased in density and frequency following fire; onlyAralia nudicaulis andQuercus alba showed statistically significant decreases in density. Overall, 29 species increased in density following fire, while 8 declined; 29 species increased in frequency, while 6 declined. However, diversity and equitability measures were depressed on the moderately burned sites, due to rapid regrowth ofKalmia latifolia. Ordination using binary discriminant analysis suggested species responded individualistically to both burning and site variation.

Comparing classification accuracy of neural networks, binary logit regression and discriminant analysis for insolvency prediction of life insurers

Past studies have documented the failure of the Insurance Regulatory Information System (IRIS) to provide adequate warning of insurer financial distress or insolvency. As a result, scholars have examined alternative parametric and non-parametric models to predict insurer insolvency. This study uses a neural network, a non-parametric alternative to past techniques, and shows how this methodology predicts insurer insolvency more effectively than parametric models.

The relation between geomorphology and forest community types of the Haut-Saint-Laurent, Quebec

The highly disturbed community types of the Haut-Saint-Laurent in southern Quebec were classified using an hierarchic approach with three levels: geomorphology, pedology and vegetation. Considered as distinct entities were, on the one hand, the herbaceous and shrub species that made up ecological groups as defined by Meilleur et al. (1992), and, on the other hand, individual tree species. Great importance was given to geomorphological variables because they are one of the most stable ecological factors of the region. We identified 47 community types from 452 sites sampled, based on a TWINSPAN analysis of the vegetation data. Binary Discriminant Analysis (BDA) showed the importance of disturbance on community composition, especially for sites included in the forest mosaic. Other important abiotic parameters included morphogenic features, topography and drainage. The BDA showed that 24 community types were found to be significantly associated with at least one morphogenic feature. The results indicate that the Beech-Hemlock association is a more stable state for mesic sites than the often cited Sugar Maple with Hickory. On glacial deposits, the presence of Thuja occidentalis is significantly associated with grazing practices. The other 23 community types are not significantly associated with any morphogenic features. On one hand, these types tend to be ubiquitous on different deposits with the same drainage class and reveal that their composition was altered through disturbances. On the other hand, some community types are more associated with microclimates as is the case of Sugar Maple with Striped Maple, or with their ecological amplitude limits as is the case of Sugar Maple with Ironwood and Aspen or Sugar Maple with Red Maple and Black Ash. A classification of the stands, given the high number of disturbed communities within the territory, appears difficult to use as a basis for ecological units.

The use of understory species as indicators of landform ecosystem type in heavily disturbed forest: an evaluation in the Haut-Saint-Laurent, Quebec

An ecological study of the Municipalit? R?gionale de Comt? (MRC) du Haut-Saint-Laurent area, a highly disturbed landscape located south-west of Montr?al (Qu?bec), was carried out. The aim of this study was to establish links between indicator species of ecosystems and g?omorphologie al features in a heavily disturbed background. Analysis to obtain the ecological groups are based on the links of herbaceous plants and shrubs. This approach is defined as a sociological one where grouping is based only on the species' affinity. In order to obtain ecological groups, similarity measures using the chi-square coefficient were calculated on the relative cover values of the 150 retained species from 452 plots. An intermediate linkage cluster analysis was then performed on the similarity matrix thus created. By this procedure, 56 ecological groups were obtained. These groups were then characterized according to abiotic variables, especially the g?omorphologie al data, by the use of binary discriminant analysis. Re sults indicate that 39 of these 56 groups were significantly associated specifically with one of 10 mor phogenic features, while 13 of these 56 groups were associated with 2 or 3 features. The ecological groups which were associated with one specific g?omorphologie al process showed a strong link to one texture and one specific drainage class. On the other hand, the groups which were associated with several features were linked with one drainage class, usually imperfect, and a certain topographic situation. However, ecological groups showing a strong association to more than one feature can be of cartographic use, provided soil texture is analysed to determine the origin of the deposit. Finally, the type of deposit becomes non determinant on vegetation where drainage and topographic variables indicate poorly or very poorly drained sites. Non-significant groups should be used as indicators of drainage and topographi cal conditions but not as indicators of g?omorphologie al deposits. The ecological groups not associated with morphogenic features demonstrated mostly a heliophilic character. Thus, the ubiquitous character of ecological groups is not exclusive to species taking advantage of disturbances.

Seed bank dynamics of Mediterranean pastures subjected to mechanical disturbance

Abstract. Vegetation and buried germinable seeds were examined in Mediterranean pastures differing both in time elapsed since last ploughing and in intensity of disturbance. Densities of buried germinable seeds in the greenhouse were determined from 10 × 10 × 3 cm soil cores taken during the autumn prior to the time of germination. Abundance of pasture species was recorded on the same sites during the following spring.The species composition of the seed bank was very similar to that of the vegetation: 80.9 % similarity (Sorensen's index). The number of species and density of buried seeds ready to germinate increased with suc‐cessional age from 0 to 4 yr and decreased later to a constant level after 9 yr. Regarding disturbance intensity, the highest species numbers are found at intermediate intensities while there is a drastic decrease at high disturbance intensity. The floristic composition of the seedbankwas analyzed through Correspondence Analysis (CA) and Binary Discriminant Analysis (BDA). Seed bank composition reflected both successional stage and intensity of disturbance. Characteristic species of both gradients were identified.

Plant-environment relationships on the Montlake wildlife area, Seattle, Washington, USA

Grassland vegetation on the Montlake fill was analyzed using TWINSPAN. Eight herb communities were recognized. Moisture, proximity to gas vents, and disturbance are the main factors that control species and community distributions. Binary discriminant analysis (BDA) and detrended correspondence analysis (DCA) were used to study species-environment relationships. BDA revealed complex species response patterns and the resultant indicator values were used to interpret the ordination axes. Species distributions are controlled primarily by moisture, but also influenced by soil pH. Multiple regressions revealed little about plant-environment relationships not discovered by BDA. Before robust nonlinear methods are available, BDA, metric ordination with data stratification and nonmetric ordination are methods that can yield satisfactory results in exploratory plant-environment studies. BDA alone is an efficient, useful first approach where response patterns of species are initially unknown.

A NUMERICAL ANALYSIS OF FLAVONOID VARIATION IN ARNICA SUBGENUS AUSTROMONTANA (ASTERACEAE)

Species of Arnica subgenus Austromontana produce a total of 23 leaf flavonoids, including simple and methylated flavone and flavonol glycosides as well as highly methylated flavone aglycones and a 6‐hydroxylated flavone. Most of the taxa exhibit considerable interpopulational variability, with the number of compounds per population ranging from 2 to 14. Analysis of flavonoid variation in 113 populations representing all 9 species of the subgenus was carried out using cluster analysis, principal components analysis, and binary discriminant analysis. Results indicate the flavonoid profile of the very rare A. viscosa is the most distinctive in the subgenus. Although exhibiting considerable interpopulational variability, all populations of A. gracilis, a hybrid taxon, form a very distinct and cohesive group, supporting its recognition at the specific level. Additionally, chemical diversification from A. cordifolia has taken place largely in the Klamath region of Oregon and California. The range of variability exhibited by A. cordifolia is reflected in these Klamath region derivatives.

Tree species distributions related to climatic patterns in Michigan

Relationships between the distributions of major forest species and climatic patterns in Michigan were examined using binary discriminant analysis ordination, Kolmogrov–Smirnov statistics, histograms, and correlation coefficients. Species occurring only in southern Lower Michigan (e.g., sassafras, Sassafrasalbidum (Nutt.) Nees) were strongly associated with warm, long growing seasons that had high night temperatures, low precipitation relative to potential evapotranspiration throughout the summer months, and low heat sums prior to last spring freeze. Species found throughout Lower Michigan but not extending into Upper Michigan (e.g., black oak, Quercusvelutina Lamarck, and white oak, Q. alba L.) had similar associations except that they were not strongly associated with low heat sums prior to last spring freeze. Conifers (e.g., balsam fir, Abiesbalsamea (L.) Miller, and white pine, Pinusstrobus L.) were associated with short growing seasons of highly variable length, high heat sums prior to last spring freeze, high precipitation relative to potential evapotranspiration even in July and August when the difference between precipitation and evapotranspiration was greatest, and low night temperatures relative to daytime temperatures. All species had individualistic correlations with major climatic statistics. Species range limits, where they occurred in the state, were generally simultaneously correlated with more than one climatic variable.

Bottomland Vegetation Distribution along Passage Creek, Virginia, in Relation to Fluvial Landforms

Persistent distribution patterns of woody vegetation within the bottomland forest of Passage Creek, Virginia, were related to fluvial landforms, channel geometry, streamflow characteristics, and sediment—size characteristics. Vegetation patterns were determined from species presence as observed in transects and traverses on landforms developed along the stream. Distinct species distributional patterns were found on four common fluvial geomorphic landforms: depositional bar, active—channel shelf, floodplain, and terrace. Independent hydrologic characteristics (flow duration and flood frequency) were determined for each of the landforms. Vegetation data were analyzed by binary discriminant analysis, principal components analysis, and detrended correspondence analysis. Results and related field observations suggest that certain species are significantly associated with specific fluvial landforms. Vegetation patterns appear to develop more as a result of hydrologic processes associated with each fluvial landform rather than from sediment—size characteristics. Flood disturbance may be an important factor in maintaining the vegetation patterns, which may therefore be used as indicators for particular hydrogeomorphic site conditions.

Vegetation ordination of subalpine meadows using adaptive strategies

The vegetation of subalpine meadows was investigated by several multivariate analytical methods. Detrended correspondence analysis (DCA) revealed overall patterns of vegetation response to environmental gradients and identified many indicator species. Binary discriminant analysis confirmed these interpretations and identified indicator species with more clarity than did DCA. A plant strategy ordination was developed from a morphological index of competitiveness and from a relative growth rate index. Results converge to these conclusions: (i) the strategy ordination was consistent with habitat types and DCA; (ii) the degree of species niche differentiation within a community varied with productivity and disturbance; (iii) morphological and physiological niche differentiation is least in productive sedge meadows and greatest in mesic grasslands where production and disturbance are moderate. The adaptation spectra of species in a community provide useful insights about site conditions, history, and response to manipulations.

Environmental controls in Pacific Northwest intertidal marsh plant communities

Relationships between environmental variables and species distribution were studied in a brackish intertidal marsh formed by the Skagit River as it enters the Puget Sound bay system in Washington. Transects were established which covered the range of environmental variation in the marsh. A grid of environmental measuring stations provided information on soil texture, organic content of soil fines, macroorganic material in the soil, soil temperatures, interstitial soil water salinity, soil redox potential, and site elevation. Binary discriminant analysis, a nonparametric method using species presence–absence data, was used to construct standardized residual matrices. Principal component analysis of standardized residuals (Q mode) indicated that salinity and soil texture were strongly correlated with the first factor generated, elevation with the second, and soil redox potential with the third. The factors explained, respectively, 48, 21, and 14% of the variance in the residuals matrix. From R-mode analysis, eight community types were derived: three dominated by Carex lyngbyei, two by Scirpus americanus, one by Scirpus maritimus, and two which are highly diverse.

Vegetation-Site Relationships in the Presettlement Forests of Northeastern Ohio

The presettlement forest of northeastern Ohio was dominated by two major forest types. The complex surficial geology and topography of the area created a variety of habitats favoring forests composed of beech and maple on some sites and oak and hickory on others. Binary discriminant analysis was applied to data from the early land survey records of Wayne, Ashland, and Holmes Counties to determine the environmental factors defining the two forest types. Beech (Fagus grandifolia) was largely confined to imperfectly drained soils derived from a relatively fine-textured till. White oak (Quercus alba) occupied the coarser-textured tills and residual soils. A two-dimensional gradient depicting, on one axis, the supply of moisture to the soil-water system by runoff and, on the other axis, the removal of moisture by internal drainage provides a framework for interpreting many of the more important species-environment relationships.

CONTROL OF VEGETATION ON CONTRASTING SUBSTRATES: HERB PATTERNS ON SERPENTINE AND SANDSTONE

Binary discriminant analysis (BDA) reveals relationships between species distributions and environmental variables. In this study, BDA was applied to transects on serpentine and sandstone in the Wenatchee Mountains, Washington. Species presence and states of ten habitat variables were recorded in each quadrat. Species response patterns significantly different from a random expectation suggested that distributions are controlled by soil moisture far more strongly on serpentine than they are on sandstone, where light and its correlates are more important. Environmental patterns were explored with direct principal components analysis (PCA) of standardized environmental variables and indirectly by PCA of the significant D‐values. The latter is biased towards variables to which species respond strongly. The results emphasize the environmental and floristic contrasts between substrates. On serpentine, direct PCA indicates that effective moisture, soil fertility, and insolation control the first three axes, while on sandstone, insolation, fertility, and effective moisture control the first three axes, respectively. The PCA of D‐values (Q‐mode) is similarly interpreted: moisture is strongly correlated with the first component on the serpentine transect, while insolation and fertility are correlated with the first component on sandstone. Species ordinations also result from Q‐mode analysis. They reflect species responses to the identified gradients and indicate mutual species‐environment interactions. These analyses conform to the hypothesis that mineral conditions of serpentine select against intolerant species and that survivors respond primarily to moisture conditions that are a result of low productivity, attributable to adverse nutrient conditions. On sandstone, the moisture gradient is less pronounced, and direct canopy effects that create variable light conditions predominate. BDA is a useful tool in ecological survey and pattern analysis. Species responses to environmental factors may be estimated quickly and directly. This method will help to focus subsequent research, improve experimental design, and generate explicit and testable hypotheses about species‐habitat interactions.

Response of Woody Species to Site Factors of Slope Angle, Rock Type, and Topographic Position in Maryland as Evaluated by Binary Discriminant Analysis

Binary discriminant analysis of 161 species lists including 134 woody taxa taken from the Maryland Piedmont and inner Coastal Plain reveals that 19% of the species respond significantly to vasiation in slope angle, 36% to rock type, and 22% to topographic position. Response to rock type was greatest because the area sampled was quite large and included extensive areas underlain by distinctive lithologies. Species groupings produced by binary discriminant analysis and based on common patterns of preference or avoidance for rock type and topographic position correspond well with associations or community types which have been recognized by previous investigators.

Binary Discriminant Analysis: A New Method for Investigating Species‐Environment Relationships

Binary discriminant analysis (BDA) is a readily understood, easily used technique for identifying binary variables and their common trends which are most important for discriminating between groups. For the plant ecologist, the technique can be used on species lists to reveal similar patterns of preference or avoidance among species responding significantly to a multistate environmental parameter such as soil type, rock type, or aspect. In Q—mode BDA, orthogonal canonical factors are obtained which represent uncorrelated floristic trends best separating the groups. Scores of species on the factors can be plotted in multidimensional hyperspace, showing how each species responds to the floristic trends. In R—mode BDA, groups of species with similar responds to the environmental parameter are identified. These groups may be interpreted as statistical associations or community components comprised of species with similar ecologies. An example using lists of woody species from the Maryland Piedmont and Coastal Plain sorted according to underlying rock type produces floristic trends which are easily interpreted and species groups which are readily understandable.