Comparative phytolith profiles of the weed grasses Echinochloa colona and Echinochloa crus-galli (Poaceae): implications for species taxonomy

Introduction Taxonomic delimitation of closely related Echinochloa species is frequently complicated by morphological plasticity and overlapping vegetative characters, necessitating complementary diagnostic approaches. This study aimed to evaluate whether integrated phytolith characteristics can provide reliable discriminatory features for distinguishing Echinochloa colona and E. crus-galli . Methods Phytolith biomineralization in E. colona and E. crus-galli was comparatively investigated by integrating epidermal silicification patterns, organ-specific phytolith morphology and occurrence, relative frequency, quantitative morphometry, silica accumulation, scanning electron microscopy (SEM), SEM–EDX elemental analysis, and multivariate statistics. Phytolith assemblages from roots, culms, leaves, and inflorescences were compared using presence–absence and morphometric data, while elemental composition was evaluated using principal component analysis. Results Distinct phytolith distribution patterns occurred on the adaxial and abaxial leaf surfaces of the two species. Across the four organs, 50 phytolith morphotypes were recorded, of which 31 occurred in E. colona and 38 in E. crus-galli , with several showing species- or organ-restricted occurrence. Hierarchical cluster analysis demonstrated substantial differentiation among organspecific assemblages, with an overall Jaccard similarity of 0.229 between the species. Similarity was greatest between corresponding leaf and inflorescence assemblages (J = 0.292 each) and lowest between roots (J = 0.053). Silicified inflorescence ornamentation also differed, with ∩- and η-type patterns in E. colona and β-type patterns in E. crus-galli . Several shared morphotypes differed significantly in relative frequency, while quantitative morphometry revealed additional size and shape variation. Discriminant function analysis of 18 shared morphotype–organ combinations showed moderate morphometric differentiation (canonical correlation = 0.612), with an apparent classification accuracy of 83.3% and leave-one-pair-out cross-validated accuracy of 66.7%; phytolith width contributed most strongly to discrimination. SEM revealed additional ultrastructural variation, while SEM–EDX detected 14 elements in E. colona and 15 in E. crus-galli . Principal component analysis of elemental composition separated the two sampled species, with the first two components explaining 97.12% of the total variance. Discussion The findings demonstrate that phytolith-based discrimination within Echinochloa is strengthened by integrating qualitative morphology, organ-specific occurrence, assemblage composition, relative frequency, morphometry, and elemental characteristics rather than relying on individual phytolith attributes. This integrative framework has potential applications in grass taxonomy, archaeobotany, and palaeoecological reconstruction, although broader populationlevel validation is required to establish the generality of the observed species-level differences.