Data Package Metadata   View Summary

To test for an association between traits and seed cold tolerance, we ran freeze tolerance trials on species with differing seed characteristics including cold stratification requirements, seed mass, seed coat thickness, seed shape, and seed maturation timing. Seeds were purchased from a seed supplier that focuses on regional seed sources (Prairie Moon Nursery, www.prairiemoon.com). We followed their cold stratification requirements for each species, which indicate the length of time seeds should be moist and cold to facilitate germination. The seed supplier determined these cold stratification requirements from a combination of experimental testing, consulting literature, and feedback from growers. Although they do not represent the absolute optimal length of cold stratification, they are stratification requirements that result in reliable germination (personal communication with seed supplier). We acknowledge that the germination requirements obtained from the seed supplier do not represent data from peer-reviewed research, but they were obtained from professionals who are familiar with growing native species and are in the business of providing reliable information to attract and maintain customers. We selected 6 species within each cold stratification requirement time (0, 30, 60, or 120 days), with the exception of 120 days where there was only one species available (Tradescantia ohiensis). Because there is likely variation in the actual length of cold stratification requirements, we split species into two categories (no required cold, moist stratification or required cold, moist stratification) for analysis purposes. The requirement for cold, moist stratification of our species was also verified (Steffen 1997) and for a few taxa the requirement was ambiguous so we did not include these species in analysis that included stratification requirements. We also selected species from different families to capture broader taxonomic coverage and to help avoid taxonomic bias. To calculate average seed mass per species, we weighed three batches of 10 seeds per species (20 seeds for species with seeds < 1 mg) and then calculated average seed mass per species. We also used seed coat thickness from an existing trait database (Orrock et al. 2023). Seed dimension measurements (length, width, and depth) were used to calculate a sphericity index and surface area to volume ratio. These metrics were calculated by assuming each seed is an ellipsoid in shape and using the three dimensions to estimate the volume and surface area of each seed. The sphericity index is the ratio of the surface area of a perfect sphere with an equal volume to the seed compared to the actual surface area of the seed (a value of 1 indicates that the seed is a perfect sphere while decreasing values indicate departures from a spherical shape). As an estimate of seed maturation timing for our species, we used seed collection data from a local land conservancy group who regularly collects and reseeds natural areas (Pleasant Valley Conservancy 2023). Exact timing of seed maturation can vary based on site characteristics and interannually variability of weather, so the seed maturation data used here are broad (e.g., months). We did not have complete trait coverage for our species (cold stratification requirement, seed mass, seed dimensions = 17 species; seed coat thickness = 12 species, seed maturation = 15 species).

To cold stratify seeds, 100 seeds per replicate (n = 6) were put in wet filter paper sealed inside a plastic bag and put in a refrigerator for the required cold stratification time for each species (30 – 120 days). Each species that required cold stratification (14 spp) and each replicate were cold-stratified separately, for a total of 84 bags. Once the cold stratification period was over, 20 seeds were placed in one of three 80mm diameter dishes lined with wet germination paper. For some species, a small amount of stratifying seeds became moldy during stratification so these seeds were not used for germination trials. Additionally, for one species (Heliopsis helianthoides) a few seeds started germinating during cold stratification, so germinated seeds were also not used for further testing. For species with no cold-stratification requirements (6 spp), dry seeds were stored in closed seed envelopes kept at room temperature and were plated as described above. Each of the 3 dishes was then exposed to a different temperature treatment (-15 °C, -26 °C, or room temperature (21 – 26 °C) control) for 2 hours, following previously published methods testing cold tolerance of various tissues for species in our system (Ladwig et al. 2018, Nettesheim et al. 2021, Henn et al. 2022). The -15 °C temperature treatment corresponded with the minimum soil temperature achieved at our field sites (Table 2). The lowest treatment temperature (-26 °C) is colder than winter soil temperatures typically observed in our region, and Following the temperature treatment, all dishes were moved to a growth chamber (12h light/dark, average day and night spring temperatures) and monitored for 2 weeks to record the number of seeds that germinated. The experimental design model was 19 species × 3 temperature treatments × 6 replicates = 342 dishes. We ran additional replicates for species with low germination in control plates. For each replicate, we calculated a log response ratio (LRR = log(germination % under treatment + 1)/germination % in control + 1) for the two treatment temperatures and used this value for further data analysis. Because we had some trials where there was zero germination in the cold treatment replicates (68 out of 279 replicates), we added 1 to the control and treatment germination percentages to allow us to include those results in analysis.

To test if cold tolerance relates to seed characteristics and treatment temperatures, we used linear mixed-effect models with seed characteristics (including seed coat thickness, mass, sphericity, surface area:volume, cold stratification requirement, maturation, and family) and cold treatment temperature, as well as their interaction, as fixed effects, species with replicate nested in species as random effects, and log response ratio as the response variable. We fit one model per trait (total of 7 models) to examine how each characteristic is related to cold tolerance. For all analysis, only replicates with sufficient baseline germination (>10% germination) in the room temperature control were included (279 out of 406 replicates, two species failed to germinate, 8 species had sufficient germination for all replicates, and 9 species had some replicates without sufficient germination). We expect that low germination in control replicates could be due to a variety of factors including not meeting other dormancy-breaking requirements, low seed viability, or slow germination.

We selected two common prairie species with dramatically different cold stratification requirements to test how winter weather may influence species differently based on cold stratification requirements. Tradescantia ohiensis requires 120 d cold stratification while Desmodium canadense has no cold stratification requirement. Both species have large native ranges and commonly occur in prairies across the Midwestern USA (GBIF 2022). Seeds were purchased from the same regional seed supplier (Prairie Moon Nursery, www.prairiemoon.com).

The same field set-up was replicated at three herbaceous-dominated sites where our target species would typically grow (Table 2). At each site, ten 1 × 1 m plots were established, 5 for snow removal and 5 as snow controls. Plot arrangement varied with site layout, but had at least 2 m buffers between plots and the edge of the grassland. Seeds of the two prairie species (T. ohiensis and D. canadense) were installed in each plot. Specifically, 50 seeds were put in a histology tissue cassette and “planted” in each plot by nestling the cassettes under the litter layer to simulate positions where newly dispersed seeds would likely overwinter. Small openings in the walls of the cassettes allowed seeds to interact with the environment while still keeping them enclosed for later collection. In early winter (December 13 - 17, 2019), 10 seed cassettes per species (20 total) were installed in each plot. All seed cassettes were tethered to fence posts with a 1 m wire to facilitate relocation and collection. After a snow event of at least 2 cm, snow was removed from snow removal plots using shovels and brooms. Snowfall varied by location, so snow removal occurred 11 times in Green Bay, 4 times in Madison, and once in Casey. Twelve data loggers (Thermochron iButtons; model DS1921G-f5#; Embedded Data Systems, Lawrenceburg, KY) that recorded temperature every 4 h were deployed at each site, with one data logger at the soil surface in each plot and two attached to fence posts at 1m under radiation shields to measure air temperature.

One seed cassette for each species per plot was collected every 2 weeks between January 15, 2020 and May 11, 2020 (with the exception of a 6-week gap in late February/early March that coincided with spring break and the onset of pandemic-related institutional shutdowns). Seeds were removed from cassettes and plated on wet blotter paper to monitor germination for two weeks. If seeds were already germinated in the cassette they were recorded as germinated seeds and included in final germination percentages but not plated. If seeds were excessively moldy in the cassette, they were not plated and also not included in the final germination percentages, since it was often unclear if seeds had germinated prior to molding. Seeds arriving moldy was mainly an issue for D. canadensis. All germination trials were run at the central site (Madison, WI) so cassettes were shipped from the other sites shortly after collection. Seed plates were checked at least 3 times per week for germination, during which germinated seeds were removed and counted. Seed plates were initially kept in a growth chamber (12h/12hlight/dark regime) and transitioned to a shelf with a grow light (12h/12h on/off regime, temperature 19 – 21 °C) in early April when the lab building closed due to the Covid-19 pandemic. For D. canadense we did not evaluate germination after the first four sampling times because seeds became moldy in later collections.

We used temperature measurements to calculate metrics that we predicted would relate to germination and cold stratification. Specifically, we calculated the mean temperature for each day at the soil surface and categorized days as cold days (mean temperature < 0 °C), cold stratification days (mean temperature 0 - 5°C), and warm days (mean temperature > 5 °C). We used the accumulated days of each type for each plot at each collection date during the study period to predict germination proportion. To test for differences in germination proportions between sites, snow treatments, and temperature through time, we used a linear mixed effects model. We logit-transformed the germination proportion because it is bounded between 0 and 1 (Warton & Hui 2011) and included site, treatment, cold accumulation days, cold stratification accumulation days, and warm accumulation days as fixed effects. We also included the interaction between site and treatment, but no interactions with temperature accumulation days. We included plot and sampling round at each site as random intercepts to account for the repeated measures from each plot.