Comparison of discriminant methods and deep learning analysis in plant taxonomy: a case study of Elatine.

Elatine is a genus in which, flower and seed characteristics are the most important diagnostic features; i.e. seed shape and the structure of its cover found to be the most reliable identification character. We used a combination of classic discriminant methods by combining with deep learning techniques to analyze seed morphometric data within 28 populations of six Elatine species from 11 countries throughout the Northern Hemisphere to compare the obtained results and then check their taxonomic classification. Our findings indicate that among the discriminant methods, Quadratic Discriminant Analysis (QDA) had the highest percentage of correct matching (mean fit-91.23%); only the deep machine learning method based on Convolutional Neural Network (CNN) was characterized by a higher match (mean fit-93.40%). The QDA method recognized the seeds of E. brochonii and E. orthosperma with 99% accuracy, and the CNN method with 100%. Other taxa, such as E. alsinastrum, E. trianda, E. californica and E. hungarica were matched with an accuracy of at least 95% (CNN). Our results indicate that the CNN obtains remarkably more accurate classifications than classic discriminant methods, and better recognizes the entire taxa pool analyzed. The least recognized species are E. macropoda and E. hexandra (88% and 78% match).

Formalized classification of ephemeral wetland vegetation (Isoëto-Nanojuncetea class) in Poland (Central Europe).

Formalized classification of the class Isoëto-Nanojuncetea has not been performed in Poland. We used 69,562 relevés stored in Polish Vegetation Database. Based on the literature and expert knowledge we selected 63 diagnostic species for the Isoëto-Nanojuncetea class. Unequivocal classification was applied in this work according to Cocktail method. A set of formal definitions was established using a combination of logical operators of total cover of species in case of high-rank syntaxa while sociological species groups and cover of particular species were used for logical formulas describing class, alliances and associations. An Expert System was prepared and applied to classify the whole data set of PVD and 1,340 relevés were organized at the class level. We stratifies the data and finally we used data set of 903 relevés to prepare synoptic tables, distribution maps and descriptions of the syntaxa. Twelve associations and two plant communities were identified. Vegetation of the Isoëto-Nanojuncetea class occur in Poland's central and southern part, with scattered stands in northern region. We described two new plant communities within Eleocharition and Radiolion alliance. The first formal classification of the Isoëto-Nanojuncetea class revealed a high diversity of ephemeral vegetation wetland found in Poland in the eastern boundary of their geographical distribution in Europe.

Rediscovery of Cyperus flavescens (Cyperaceae) on the northeast periphery of its range in Europe.

In recent years, three large populations of Cyperus flavescens were found in Poland, the richest occurrence of this species in over 30 years. The goal of this research is to determine the habitat factors lead to the mass occurrence of C. flavescens and the present situation of that species and its habitat in Central Europe. Soil conditions of the three populations were studied. To determine the correlation between the occurrence and abundance of species and the chemical parameters of the soil, the DCA and CCA methods were used. The DCA of environmental Ellenberg values was made for all relevés known from Poland. The occurrence of C. flavescens in plant communities in Central Europe was studied. The maximum entropy method was used for potential distribution analysis of C. flavescens. All analyzed traits are important for this species and none has an advantage over another, so the environmental factor affecting the occurrence of C. flavescens is different from the tested. Analysis on Ellenberg values indicate that the longest gradients are temperature, moisture and nutrients. The analysis of vegetation data involving Cyperus flavescens available from Central Europe indicates that this species occurs mainly in the company of Juncus bufonius and Plantago intermedia, whereas other species of the Isoëto-Nanojuncetea class appear rarely. In MaxEnt analysis based on bioclimatic variables, the most important variable is BIO1 (Annual Mean Temperature). The results of our observation indicate that anthropogenic factors such as grazing livestock have a positive effect on the occurrence of the species. It is also very likely that the species is promoted by very warm summers with only short periods of heavy rains. A map of the potential distribution of C. flavescens in Central Europe created according historical and future data show an extension of the range of potential habitats to the north and east.

Seed morphometric characteristics of European species of Elatine (Elatinaceae).

Elatine L. contains ca. 25 small, herbaceous, annual species distributed in ephemeral waters in both hemispheres. All species are amphibious and characterized by a high degree of morphological variability. The importance of seed morphology in Elatine taxonomy has been emphasized by many authors. The degree of seed curvature and seed coat reticulation have been traditionally considered very important in recognizing individual species of this genus. Seed morphometric characteristics of 10 Elatine species, including all European native taxa, are provided on the basis of material from two or three populations of each species. A total of 24-50 seeds were studied from each population, altogether 1,260 images were used for the morphometric study. In total, six parameters were measured from SEM pictures: object surface area, profile specific perimeter (object circuit), rectangle of the object (a) length, rectangle of the object (b) width, angle of the seed curvature, and number of pits in the seed coat counted in the middle row. Our study shows that the range of morphological variation of seeds in European species of Elatine is great, both between the species and the populations. Discrimination analysis showed that all six traits significantly differentiate the populations studied (λ = 0.001, p < 0.001), and the greatest contributions were "number of pits", "rectangle_a", and "the angle curvature". Multidimensional scaling based on a correlation matrix of Mahalanobis distance of the six features studied revealed the greatest similarity between the three populations of E. alsinastrum, E. macropoda, and E. hexandra. Regarding interspecific differences, a Kruskal-Wallis tests showed that, in many cases, lack of statistically significant differences between species relative to the studied seed traits. If distinction of species is only based on seeds, especially if only a few seeds are evaluated, the following species pairs can be easily confused: E. alsinastrum and E. orthosperma, E. hexandra and E. macropoda, E. campylosperma and E. hydropiper, as well and E. gussonei and E. hungarica. We found no diversity in seed coat micromorphology within pits that could have potential taxonomic importance. An identification key and descriptions of species are provided on the basis of seeds traits.