Introduction
The morphological description of seeds and diaspores offers essential
information for scientists and practitioners in a wide variety of
fields, including botany, restoration, conservation, ethnobotany,
archaeology, and agriculture. Diaspore traits, such as size, shape,
colour, surface structures, and the presence of appendages are needed to
establish the identity of particular diaspores that become detached of
their mother plant (Martin and Barkley 1961), for instance in seed lots,
seed traps, soil seed bank, archaeological sites, or forensic
investigations. Moreover, integrating diaspore morphological traits into
theoretical plant regeneration framework can lead to major advances in
predictive evolutionary and ecological models, and thereby support
conservation and restoration actions (Saatkamp et al. 2019).
Throughout the years, the demand for knowledge of diaspore morphology
has led to numerous compilations of text descriptions and/or images of
diaspores in books, guides and atlases (e.g., Martin and Barkley 1961,
Brouwer and Stählin 1975, Beijerinck 1976, Sweedman and Merritt 2006,
Bojňanský and Fargašová 2007, Cappers et al. 2012). In the last two
decades, databases have been built to synthesise and centralise
information on diaspore traits (e.g., Kleyer et al. 2008, Hintze et al.
2013, GEVES 2022, Royal Botanic Gardens Kew 2022), facilitating large
scale analyses. Along with databases, standardized protocols were
established for trait measurements to allow for the integration of data
with different sources. These included methods for the description of
diaspores, which consist of the quantification of size and other
morphometric measurements (most reported as taxa mean or range values),
and the classification of attributes either based on visual (perceptual)
categories and/or functional structures and/or anatomical parts
(Römermann et al. 2005).
Recently, the pressing need for new solutions to deal with environmental
crises, together with the surge in applications of machine learning and
image analysis in ecology and evolution, calls for an upgrade of the
diaspore morphological datasets. The automated extraction of information
from digital images provides the opportunity to collect quantitative
phenotypic data in large quantities, enabling the investigation of high
dimensional and complex relationships between traits and their
interaction with environmental variables (Lürig et al. 2021).
Furthermore, the use of machine learning algorithms to classify images
and/or suites of traits can allow for the automation of taxa
identification, making the task faster and not exclusively dependent on
experienced taxonomists (Borowiec et al. 2022, Loddo et al. 2022).
Here, we present DiasMorph, a comprehensive dataset of morphological
traits and images of diaspores from Central Europe. It provides images
of 94,214 diaspores from 1,437 taxa in 513 genera, and 96 families,
captured with a standardized and reproducible method (Dayrell et al.submitted ). The dataset also compiles information on quantitative
morphological traits extracted from the images following an image
analysis method and include not only traditional morphometric
measurements, but also colour, and shape features made available for the
first time in a large dataset (Dayrell et al. submitted ). The
quantitative traits records correspond to measurements of individual
diaspores, an input currently unavailable in trait databases that will
allow for several approaches to be used for a complete exploration of
the morphological traits of these species. We also included information
on the presence and absence of appendages and structures in the
diaspores of the evaluated taxa. By making these data available, we aim
to encourage initiatives to advance on new tools for diaspore
identification, further our understanding of morphological traits
functions, and provide means for the continuous development of image
analyses applications.