Seed banks, the collection of viable seeds in the soil, are particularly important determinants of population survival in highly variable environments. Predictions of increased stochasticity in the amount and timing of precipitation in desert environments raise the question of how seed banks of desert species will respond to climate change, and ultimately, whether these species will persist. Here, we present data from our long-term studies of germination requirements and seed bank dynamics in a rare desert gypsophile perennial, Arctomecon californica (Las Vegas bearpoppy). Arctomecon californica is a relatively short-lived plant that recruits from seed in sequences of unusually favorable years. We used germination experiments, an in situ seed bank study, and a 15-year field seed retrieval study to examine factors affecting seed bank persistence. In the germination study, a majority of seeds remained dormant, despite a wide variety of treatments, suggesting that a large proportion of the seed dispersed each year has cue-nonresponsive dormancy. Our seed bank study showed that seed density varied widely between sites, among transects, and among samples within a transect. The patchiness of seeds in the soil highlights the importance of protecting large areas where A. californica populations are known to have existed in the past. The seed retrieval study provided strong evidence that this species has a long-lived seed bank in which only a small fraction of seeds (roughly 5%) become nondormant each year, allowing seed banks of this species to last up to 20 years without a seed production event. Whether this impressive life history strategy can maintain the species in the face of climate change depends on the future frequency of the well-timed precipitation that allows for establishment of new cohorts of adult plants.

Study organism and habitat

Arctomecon californica, the Las Vegas bearpoppy, is a rare herbaceous perennial plant of the Mojave Desert.  The species is restricted to barren ‘badlands’ soils high in gypsum or with other unusual chemical attributes. It has lost much of its habitat, primarily due to urbanization, and it is limited to fragmented, remnant populations in the Las Vegas Valley, larger populations around Lake Mead in Nevada and Arizona, and a disjunct population in the Lower Grand Canyon. The species was recently petitioned for listing under the Endangered Species Act.

We chose two sites in different parts of the species range for the field experiments reported here. The first site was a small Las Vegas bearpoppy preserve at the northeastern corner of the North Las Vegas Air Terminal in the Las Vegas Valley, while the second site was within a bearpoppy population on the road to the St. Thomas historic townsite near the north end of the Lake Mead National Recreation Area.

Arctomecon californica is a relatively short-lived plant (known maximum life span of seven years) that recruits from seed in sequences of unusually favorable years. It has been thought to depend on a long-lived seed bank for population persistence through sometimes long periods unfavorable for recruitment and survival.

Seeds of A. californica are known to exhibit morphophysiological dormancy (i.e., the embryo is immature at seed dispersal and also requires a physiological cue for germination). Wide variation among species exists in the conditions required to break morphophysiological dormancy. In some species, warm conditions break the physiological dormancy of the embryo, and embryo growth occurs at cold temperatures; however, some seeds with this morphophysiological dormancy can complete both dormancy break and embryo growth during moist chilling. Germination of nondormant seeds in A. californica depends on winter/spring precipitation, which is highly variable among years.

Seed germination study

The seed germination study was carried out in 1995 with seeds collected from ripe capsules at the North Las Vegas Air Terminal (NLVAT), at Saint Thomas Road (STHOM), and at Rainbow Gardens (RBG) on BLM land in June 1995. The germination experiment was carried out in the laboratory by placing recently harvested seeds in Petri dishes on the surface of water-saturated germination blotters, with subsequent incubation in temperature-controlled chambers. The original experimental design included a total of 1032 experimental units (Petri dishes) and 462 treatment combinations. Each treatment included either four replicate dishes of 25 seeds or two replicate dishes of 50 seeds as detailed below.

We determined in earlier experiments with a single 1985 seed population from Rainbow Gardens that a single warm moist stratification at 20°C plus a single short to intermediate cold stratification (4–12 wks) at 2°C did not result in significant dormancy break for A. californica seeds previously stored under laboratory conditions for three years (germination ≤2%). For this reason, we did not include a formal warm moist stratification pretreatment in the 1995 experimental design. Our treatments included three seed populations (NLVAT, RBG, and STHOM), a series of 11 dry storage temperature by duration treatments, seven chilling duration by temperature combinations, and two post-chilling incubation temperatures.  Our initial design included a treatment that included high-temperature dry-after-ripening (50°C) followed by moist chilling on the logic that this combination was most likely to result in first-year dormancy break in this spring-emerging species.

The laboratory seed germination experiment was carried out in two steps: an original experiment (data not shown), and an additional experiment that added a re-chill treatment as a continuation. The original experimental design was factorial, with three seed populations (accessions), 11 storage duration-temperature combinations, four chilling durations (0, 4, 8, and 12 weeks), two chilling temperatures (1°C, 5°C) and two post-chilling incubation temperatures (5/15°C or 10/20°C for 12h:12h). Treatment combinations included either four Petri dish replicates (early treatments) or two replicates (remaining treatments).

For the storage treatments, seeds were stored at 20, 30, 40, and 50°C under dry conditions for 0 to 18 weeks. Shorter maximum durations were used at the two warmer temperatures based on data from other species showing that prolonged storage at constant high temperature could damage the seeds. Storage duration and storage temperature were therefore not combined in a full factorial design. After each chilling treatment, seeds were incubated at one of the two incubation temperatures for 2 weeks and scored for germination. In the earliest treatments, seeds were then immediately subjected to viability evaluation; viability was consistently extremely high (≥99%).

After observing essentially no germination in any of the initial treatments, we decided to subject the later treatments to a second chilling experience after the 2-week germination period. We chose a 12-week cold stratification period for the re-chilling treatment. This likely exceeds the maximum chilling duration that would be experienced in the field. After this second 12-week cold stratification at the same temperatures as the original chilling treatments, the seeds were again incubated at the two original incubation temperatures for 2 weeks and scored for germination and viability. This second round of experimental treatments resulted in some dormancy break as reported below.

We have found that long-term viability in dry storage is usually a necessary but not sufficient indicator of the ability of seeds to persist across years in field seed banks. We evaluated long-term viability ex post facto for the seed lots used in germination and retrieval studies reported here by performing tetrazolium tests on four replicates of 50 seeds for each seed lot following 27 years of unsealed laboratory storage at room temperature (ca. 20°C). Seeds lose viability gradually over time in dry storage and this is reflected in a gradual diminution in respiration and resulting red color after tetrazolium staining. As individual seeds lose viability at different rates, this results in a range of embryo color intensity. We took this into account by scoring seeds in three categories; viable (red or dark pink), marginal (pale pink), or nonviable (white). Seeds in the marginal category would probably not be able to produce seedlings, but the fact that they are still respiring indicates that they maintained viability longer than seeds in the nonviable category.

In situ seed bank study

Plots were established in bearpoppy habitat at NLVAT and at STHOM in March 2005. Seed bank plots were set up within seedling establishment transects that were part of a separate study, therefore, seedling data are not presented here. Seedling emergence transects were located within areas where living or dead adult bearpoppies and gypsum soils were visible.

At NLVAT, 10 transects used for seed bank sampling were established across the study site. Quadrats were located at distances along the transect lines determined by random numbers generated using a stopwatch and were designated only if seedlings were found within the randomly located 4 dm² quadrat. At STHOM, the search protocol was different due to lower seedling densities. The entire area within 2 m on one side of each of five transects was searched for seedlings. When seedlings were located, 4 dm² quadrats were designated around them, and these plots were used for both the seedling study and the seed bank study.  The transects were laid out in parallel at each site, but spacing and length of transects at each site were constrained by the requirement that there be suitable soils and evidence of bearpoppy occupation within each transect. This resulted in variable inter-transect spacing and transect length. Paired seed bank samples were taken just outside of each seedling quadrat using a cylindrical soil can 6 cm in diameter and 4 cm in depth. Samples were lifted into labeled paper collection bags with the aid of a mason trowel. Sample size was 247 for NLVAT and 294 for STHOM. Samples were returned to the lab for sieving and seed extraction. It was relatively easy to visually identify the 2 to 3 mm-long jet black seeds of A. californica in the white soils where they occur. Seeds in each sample were counted and tested for viability using tetrazolium staining.

Seed retrieval study

Our seed retrieval study was conducted with seeds collected from ripe capsules from the same three populations described above, at two sites, over 15 years (1995–2010). Seeds were separated into sets of approximately 100 using a volume estimate and placed along with identification tags in 3x3 cm flat packets made of nylon mesh, which were folded over and stapled. The seeds were returned to the field within 3 days of harvest. The installations were carried out at NLVAT and STHOM. For each replicate, packets of each seed collection were buried as a set at approximately 1 cm depth by lifting and then replacing the surface lichen crust with a mason trowel. Packet groups were installed in three linear blocks at each site with 20 packet sets at approximately 0.5 m spacing per block.

Starting in early summer 1996, seed packets at NLVAT were retrieved yearly through 2010, for a total of 15 retrieval dates. Retrievals at STHOM were discontinued in 2004, when ground squirrels destroyed the seed packets. On each retrieval date except for the final date, one set of seed packets from each block was excavated and taken to the laboratory for processing. On the final date, all remaining seed packets at NLVAT were retrieved. For each packet, any recently germinated seeds were counted and removed, and remaining seeds were incubated at a near-optimum temperature (5/15 °C) for 4 weeks. Seeds that germinated during this period were scored as germinable (i.e., non-dormant). Seeds remaining ungerminated at the end of the incubation period were then evaluated for viability using tetrazolium and were scored as nonviable, viable and dormant, or empty and presumed germinated in the field.