DiaCurv: a value-based curvature analysis application in diatom taxonomy

Abstract

DiaCurv is a software application that uses a simple approach to calculating curvature by identifying the chord (length) and the sagitta (length) from a mid-point of an object or part of an object. This software application and associated R scripts provide the users with the possibilities to identify and apply curvature analysis in diatom taxonomy. Different taxa from the genera Cymbella and Luticola, with good size diminution series, were selected to validate curvature metrics for shape evaluations in taxonomy. The dorsal ventral curvature of Cymbella makes species within this genus a natural choice for comparison. In contrast, the lanceolate, elliptic-lanceolate to rhombic forms of species within Luticola are more problematic. Machine-learning techniques (principal component analysis and linear discriminant analysis (LDA)) were used to evaluate the curvature metrics for distinguishing diatom taxa. Selected taxa (C. aspera, C. cymbiformis, C. diversa, C. excisa, C. japonica, C. neogena, C. neuquina, C. parva, C. saxoicola, C. tumidula) were identified by valve curvature, except when valve shape was not a distinguishing character for the species. With the lanceolate, elliptic-lanceolate shapes of Luticola, curvature was effective in separating the linear-lanceolate forms (e.g., L. australomutica, L. goeppertiana, L. murrayi, L. saprophila), and some broadly lanceolate forms (L. crozetensis, L. subcrozetensis) but not elliptic-lanceolate forms within the same size range (e.g., L. katkae, L. pseudocharcotii, L. yellowstonensis). The experimental results indicated that LDA was more suitable for distinguishing taxa. Valve shapes which are circular, or centrally constricted or with polar differentiation or 3D in complexity may not be appropriate for curvature analysis unless specific parts of the valve are selected. The merits of this approach, or any other type of shape analysis, are that the subjective interpretation of shape differences between taxa is removed or reduced.

Keywords:

• diatom
• curvature analysis
• Cymbella, Luticola, multivariate analysis

Acknowledgements

The authors thank Koeltz Scientific publishers and Prof. Zlatko Levkov for approving the use of published images and associated data from the book series Diatoms of Europe, Diatoms of the European Inland Waters and Comparable Habitats. The authors state here that there is no conflict of interest with this work, including financial interest with the availability of the programme and related statistical functions.

Supplemental data

Supplemental data for this article can be accessed http://dx.doi.org/10.1080/0269249X.2017.1368718.

Additional information

Funding

The support for this research was supplied through a Research Advisory Council grant (RAC 2014–2016) to PBH from the Canadian Museum of Nature.