Most terrestrial plants disperse by seeds, yet the relationship between seed mass, seed dispersal traits, and plant dispersion is poorly understood.  We quantified seed traits for 48 species of native and introduced plants from grasslands of western Montana, USA, to investigate the relationships between seed traits and plant dispersion patterns.  Additionally, because the linkage between dispersal traits and dispersion patterns might be stronger for actively dispersing species, we compared these patterns between native and introduced plants. Finally, we evaluated the efficacy of a global trait database, the TRY plant traits database, versus locally collected data for examining these questions.

This archive contains species-level data used in analyses, including species metadata (origin, growth form), mean values of measured seed traits (size metrics and type of dispersal structures), two metrics of dispersion (local and broad scales, respectively) derived from grassland surveys in the study region, and information on the seed mass accessed from the TRY traits database. Note that the latter seed mass data could not be included in the archive, but can be acquired directly from the TRY plant traits database (https://www.try-db.org/TryWeb/Home.php).

Our study took place in semi-arid grasslands of the Intermountain Region in western Montana, U.S.A. The native system is dominated primarily by bluebunch wheatgrass (Pseudoroegneria spicata) with other grasses and a great variety of forbs diversifying the system, but it is heavily invaded by exotics. We identified our study species, comprised of 23 native and 25 exotic species, to reflect a range of dispersion patterns by using data from 620 1-m² vegetation plots from 31 grassland sites spread over 20,000 km² of western Montana. Plant dispersion patterns were defined at a local-scale by the proportion of plots occupied within a site and at a broad-scale by the proportion of sites occupied per species.

For each species, we collected at least 50 seeds from each of 10 plants at each of 3 locations in Missoula and Lake County, Montana in either 2020 or 2021. Collection locations were chosen opportunistically based on species presence and hence differed by species. Although these locations did not align with sites surveyed for species dispersion per se, they were generally drawn from the central portion of the study area. Seeds were stored in a laboratory under ambient conditions until measurements were taken, at which point they were cleaned by hand and sorted based primarily on visual characteristics to remove potentially non-viable seeds. To determine the mean seed mass per species, we weighed a fixed number of samples (three or four) from each of the three locations. The number of seeds weighed per sample was set per species to ensure a total mass >1.5 mg, the minimum reading needed for an accuracy of 2% per the specifications of the balance. For 32 of our 48 species, only 10 seeds were needed to reach this minimum. For remaining species, we increased the number of seeds included per sample in increments of 10 (range 20–150 seeds/sample) until the minimum mass was reached. Seed mass included the entire diaspore (e.g., endosperm, seed coat, awns, and dispersal appendages) to ensure that all species could be treated in the same way (e.g., dispersal appendages such as wings would have been very difficult to remove from small-seeded species). Though the inclusion of dispersal appendages potentially biases seed mass estimates for this subset of species, we note that this bias should be small relative to the large variation in seed mass across species. Indeed, estimates for three exotic species (Lactuca serriola, Taraxacum officinale, and Tragopogon dubius) with pappuses showed that these structures increased seed mass measures by <12%. For the remaining measurements, we used a ProgRes C10 camera (Jenoptik, CCD/CMOS) to create images of 20 seeds per species drawn from the 3 sampling locations (n=6 from two locations and n=8 from the third, chosen randomly). We used the images to obtain the following measurements for each seed via ImageJ software (Rasband 1997-2018): seed length (maximum), seed width (maximum), and seed surface area. These seed measurements excluded dispersal structures. Mean values per species for all seed size measurements are included in the species-level dataset archived here.  Finally, we inspected seeds to determine whether seeds of each species possessed dispersal structures including pappuses, awns, wings, or plumes. For smaller-seeded species, we accomplished this using the seed images and also checked the literature to assure that dispersal structures were not missed.

To enable comparison of empirical seed measures to those available in online trait databases, we used the TRY plant trait database (accessed 22 September – 7 October 2022), a global database integrating ~700 datasets including other major collective databases. This database included seed mass data for 44 of our 48 species but contained insufficient data to evaluate the other seed traits (i.e., length, width, and surface area) we measured (i.e., for only 2-40% of our study species). Importantly, 63% of n=831 seed mass records obtained from the TRY database could not be used in analyses. This is because these contained duplicate data that resulted from the consolidation of many datasets with common sources. See the publication for a full description of our process for identifying duplicate values. Remaining seed mass values from the TRY database were averaged to generate the mean estimate used in analyses. See the archive for sample size information per species.

See the README tab within (or the README.md file included in this archive) for detailed information about each data field included.