Seeds are the colonizing propagules for many plants and may therefore contribute to the filtering of species during the process of colonization and community assembly. Environmental filtering of seed traits may occur among species and influence community composition, or within species and influence the environmental breadth that a given species inhabits. To test for evidence of such filtering of seed traits, we measured morphological and germination traits of seeds of 408 angiosperm species collected across an elevational gradient in the eastern Tibetan Plateau grasslands. We tested for elevational filtering of traits at the species level, as well as within 22 of those species that occurred at different elevations, in order to test whether within-species variation reflected among-species patterns. Elevational patterning occurred in both seed morphology and seed germination. Seeds were smaller, more elongated, and had a higher surface area:volume ratio and shorter germination times at higher elevation. Seed morphology was associated with germination such that more elongated and smaller seeds with a higher surface area:volume ratio germinated faster, leading to earlier germination in seeds from high elevation. Within species, elevational variation in seed traits was observed in several species, but species differed in how those traits were distributed across elevation. These results suggest that taxonomic differences in seed traits may contribute to elevational variation in the species composition of plant communities, but that seed traits may be variably selected by elevation within species.

Study region and sites

The study region included the Gannan and Ruoergai located on the north-eastern edge of the Tibetan Plateau in China (100°45'55"–104°45'30"E, 33°35'06"– 35°35'16"N), about 55000 km² . The climate is mainly alpine with a mean annual precipitation of 532-693 mm. The growing season generally ranges from late April to early November, but only the period from June to August (about 90 days) is free of night-time frost. Mean annual temperature across all sites is 2.2°C (-18.6°C on January and 18.0°C in July).High-elevation sites had lower temperature and higher precipitation than low-elevation sites. The habitat type is primarily alpine meadow, which is dominated by native monocotyledons (such as Poaceae, Cyperaceae) and native dicotyledons (such as Asteraceae, Brassicaceae, Fabaceae, Gentianaceae, Lamiaceae, Ranunculaceae, Scrophulariaceae).

We established 86 study sites across this region at elevations ranging from 1718 m to 4070 m. The sites were distributed along several mountain ranges within areas with low grazing disturbance. The area of each species’ collection site was at least 30,000 m², in order to obtain at least 1,000 seeds for each species. The elevation of each site was recorded with a GPS instrument.

Seed collection

Across all sites, we identified 408 unique species, most of which were annual and perennial herbaceous plants, but a few were perennial woody plants . 375 species were found in only one elevation, 32 species were found in two elevations, 1 species was found in three elevations, for a total of 442 species x site combinations. Perennials were more prevalent at high elevation than annuals. These species were members of 54 plant families, following the Angiosperm Phylogeny Group Ⅳ (2016). 

Within each site, at least 1000 ripe seeds of each species were collected from at least 20 distinct individual plants of each species in late summer and autumn of 2012. Plants were haphazardly sampled to represent an even distribution across the area of the study site. Seeds were cleaned of appendages (mostly pappus) and stored in paper bags. After air-drying to a constant mass at room temperature (15°C), one-third of the seeds were randomly selected for measurements of morphological traits. The remaining seeds were stored dry at the Research Station of Alpine Meadow and Wetland Ecosystems of Lanzhou University (elevation 2950 m a.s.l, without temperature regulation), the storage temperature was similar to the ambient alpine temperatures. Seeds were stored from the time of collection (late summer and autumn 2012) to the beginning of the germination season in the study area (April 2013). In many alpine and sub-alpine species, dry storage can reduce dormancy and promote germination.

Seed traits

We measured and assessed three morphological traits of seeds: mass, shape, and surface area:volume ratio (SVR). To measure mass, three random samples of 100 seeds were weighed with an analytical balance from each species from each site. The mean of the three measurements, divided by 100, was used as the estimate of individual seed mass that was for analyses.

To measure surface area, 10 randomly selected seeds from each species and site were scanned, and the image was analyzed with Epson Expression 10000XL and imaging software (WinSEEDLE, Canada). The depth of each seed was measured using digital Vernier calipers to enable calculations of seed volume and surface area. The shape of each seed was calculated as the variance of seed dimensions (length, width, and depth), after transforming values so that length is unity. With this metric, perfectly spherical seeds have a value of 0, and elongated seeds have values of up to 0.3. 

Germination experiment

The germination experiment was conducted at the beginning of April in 2013. 50 seeds of a given species and site were incubated on a double layer of filter paper moistened with distilled water in Petri dishes (9cm diameter), with three replicate plates per species (150 seeds per species x site combination). The Petri dishes were placed in temperature chambers with a diurnal fluctuation of temperature (25°C day, 12 h light; 5°C night, 12 h darkness) with a 12-h light and 12-h darkness and a relative humidity of about 70% to reflect natural seasonal temperature fluctuations. This temperature regime was chosen to encompass the temperature range that occurs during the germination period across all elevations. The average temperature decreases with increasing elevation, but the onset of the germination season is later at higher elevation. As a consequence, the temperature range during the germination season are similar at different elevations. The experimental temperature range (5/25°C) was selected to ensure that all species experience a temperature range that occurs within the location of their collection.

Each day, the germinated seeds (radicle protruding from the seed coat) were recorded and removed from the Petri dishes. Distilled water was added to the filter paper as needed. The seed germination experiment lasted 40 days, after which negligible additional germination was observed. If seeds did not germinate by 40 days, the germination time was recorded as a missing value. The viability of un-germinated seeds was tested using the triphenyltetrazolium chloride test (TTC) . Germination traits were calculated using only viable seeds.

Two germination indices were used to quantify germination on each plate: Germination Proportion (GP) was measured as:

GP = n/N

where n = the total number of germinated seeds in a plate, and N = total number of viable seeds in the plate. Germination time (GT) was measured as:

GT =∑ (G_i×i) / ∑ G_i

where: i = number of days between seed sowing (day 0) and seed germination; G_i, number of seeds germinated on day i.