In October 2020, mature P. kansuensis seeds were collected from five natural populations on the Qinghai–Tibet Plateau ( Table 1 ). Data on the collecting sites’ mean annual temperatures and precipitation were obtained from nearby weather stations. According to Hu et al.’s (2015) and Zhang et al.’s (2020) methods, seeds originating from the Tibet Plateau (SN and LZ) and the Qinghai Plateau (YS, GL, and HB) were classified into cool and warm habitats, respectively, as the habitats’ mean annual precipitation and temperatures were distinctly different ( Table 1 ). As the mature seeds were stored in capsules before dehiscence (Sui et al., 2016), several hundred individuals containing indehiscent capsules were collected from each of the five collection sites and taken to the laboratory. The seeds were gently shaken out of the cracked capsules. Shedding seeds were cleaned using a sieve to remove the residues. As P. kansuensis seeds have non-deep physiological dormancy (Hu et al., 2022), and previous studies have suggested that dry storage at room temperature is an effective method for breaking this dormancy type (Liu et al., 2011; Baskin and Baskin, 2014). Therefore, the seeds were stored in an envelope at room temperature (20%–30% relative humidity, 10−20°C) to after-ripen until germination tests were conducted.

The seed size (including length and width) of 50 seeds originating from different populations was measured, and their thousand seed weight was measured in five sub-samples of the five seed origins. The thousand seed weight was determined in accordance with the International Rules of Seed Testing (ISTA, 2009).

In March 2021, five replicates of 50 seeds originating from different populations were placed on two sheets of filter paper (Jiaojie, Fushun, China) in 9-cm-diameter Petri dishes and moistened with 6 mL of distilled water. The seeds were incubated in an incubator at 10, 15, 20, 25, 30, and 35°C (Top Instrument, incubator model RTOP−260Y), with a 12 h light (mean photon flux density of 60 μmol·m^-2·s^-1, 400−700 nm) and 12 h dark diurnal cycle. The number of germinated seeds was examined under white fluorescent light and monitored every 6, 12, 18, and 24 h until no further germination was observed within three days. The criterion for germination was the length of the radicle reaching at least 2 mm. The germinated seeds were removed from the Petri dishes. To avoid the variance caused by incubator conditions, each incubator’s temperature, humidity, and light were monitored daily, and their Petri dishes were randomly rearranged every second day.

The seeds’ germination responses to water potential were determined by incubating the seeds in light at 20°C, with a water potential of 0, −0.2, −0.4, and −0.6 MPa. The germination substrates’ water potential was determined using polyethylene glycol 6000 (PEG) solutions, which were prepared according to Michel and Kaufmann’s (1973) methods. For each treatment, five replicates of 50 seeds were planted in 9-cm-diameter Petri dishes on two layers of filter paper and moistened with 5 mL of distilled water (control) or different concentrations of PEG solution. The Petri dishes were sealed in Parafilm to reduce evaporation. To ensure relatively constant water potential during the germination period, the seeds were transferred to a new filter paper with fresh solution or distilled water every two days. The number of geminated seeds was recorded every 6, 12, 18, and 24 h until no further germination was observed within three days. Seeds with a radicle length exceeding 2 mm were considered germinated.

A two-way analysis of variance (ANOVA) was used to examine the effects of temperature or water potential, seed populations, and their interactive effects of seed populations and temperature or water potential on germination percentage and rate (1/T ₅₀). The significant difference in seed germination percentage, germination rate, cardinal temperature (T _b, T _o, and T _c), and base median water potential [Ψ _b(50)] from the same origin under different temperatures or water potentials was tested with a post-hoc test with multiple comparisons of the means, following a one-way ANOVA. The germination percentage data were standardized and transformed using arcsine transformation to meet the assumption of a normal distribution and homogeneity before analysis. All data were processed with IBM Statistical Product and Service Solution (SPSS) software (version 19.0, Shanghai, China). For a given temperature or water potential, the cumulative germination percentage for each replicate of seeds from different collecting sites was probit-transformed and regressed against time, and the time for cumulative germination (t _g) to reach different percentiles (20%−80%) was estimated by the function described by Steinmaus et al. (2000).

The temperature range for seed germination was separated into suboptimal and supraoptimal temperatures in the thermal time analysis, and the temperature with the highest germination rate (1/T ₅₀) was considered the demarcation of the sub- and supraoptimal temperature ranges. Each subpopulation’s germination rate (germination percentage 20%−80%) was regressed against the temperatures. Linear models (Eqs. 1 and 2) were employed to estimate the cardinal temperature (T _b, T _o, and T _c), as suggested by Hu et al. (2015).

where t _g is the actual time to germination for a given percentage, T is the actual germination temperature, T _b is the base temperature, below which seeds will not germinate, T _c is the ceiling temperature, above which seeds will also not germinate, and θ ₁ and θ ₂ are the thermal times of the suboptimal or supraoptimal temperature, respectively.

For the thermal time model construction in the suboptimal or supraoptimal temperature ranges, the cumulative germination values [probit (g)] from all monitoring times and suboptimal or supraoptimal temperatures were pooled and regressed against a function of time (t _g) and temperature (T), according to Eqs. 3 and 4 (see Hu et al., 2015).

where θ _T is the thermal time constant for all individual seeds, σ _Tc is the standard deviation among individual seeds in the population or the inverse of the slope of the probit regression line, T_{c (50)} is the median temperature for germination, and the value of T_{c (50)} can be calculated according to the regression of time when g = 50% or probit (g) = 0.

The hydrotime models (Eqs. 5 and 6) were used to estimate the hydrotime constant θ _H (MPa−days) and the base water potential Ψ _b(g) (MPa), as described by previous studies (Bradford, 1999).

The traits of seed germination strongly depend on the environmental conditions in which the maternal plants grow (Donohue et al., 2005). Temperature, in particular, is considered a crucial abiotic factor in determining seed germination, species distribution, and community composition (Fenner and Thompson, 2005; Cochrane et al., 2014). Species originating from tropical regions always require a higher T _b for seed germination than those from temperate regions (Trudgill et al., 2000). Dürr et al. (2015) concluded that crop species of tropical origins have the highest T _b, whereas wild species and trees originating from cool growing regions have the lowest T _b. Similar conclusions have been drawn for the same genus from different natural populations (Steinmaus et al., 2000; Hu et al., 2015; Bao et al., 2019; Zhang et al., 2020). For example, legume or Stipa species collected from the Qinghai–Tibet Plateau have a lower T _b for germination than those collected from the Alax Desert and Loess Plateau (Hu et al., 2015; Zhang et al., 2020). Consistent with these findings for the same genus, one species from cool habitats can have a lower cardinal temperature for seed germination than the same species from warm habitats. For example, Bao et al. (2019) reported that Achnatherum inebrians seeds from cool habitats had a lower T _b, T _o, and T _c for seed germination than those from warm habitats. In the present study, although the P. kansuensis seeds all originated from populations inhabiting the Qinghai–Tibet Plateau, the seeds from warm habitats (SN and LZ) had a higher T _b for germination than the seeds from cool habitats (HB, YS, and GL). By contrast, P. kansuensis from the Tianshan Mountains of Northwest China can germinate at 5°C (Hu et al., 2022). This does not conflict with the present study’s findings, as the annual average temperature of P. kansuensis originating from the Tianshan Mountains (−4.8°C) was lower than that of the populations in our study (2.1°C). Thus, none of the P. kansuensis seeds from the Qinghai–Tibet Plateau germinated at 5°C. A possible explanation for this variation is that species distributed across wide geographical regions often demonstrate relatively large variations in germination characteristics according to their provenance (Imbert et al., 1999; Fenner and Thompson, 2005). Imbert et al. (1999) and Baskin and Baskin (2014) suggest that temperature during the seed maturation of one species from different populations, has a local adaptive effect on the offspring germination characteristics of seeds (Imbert et al., 1999) because the minimum temperature required for germination provides an adaptive benefit to prevent premature or overdue germination during the Qinghai–Tibet Plateau’s short summer (Hu et al., 2015). Furthermore, the adaptive strategy of germination in response to habitat temperature also largely determines the destiny of the recruited seedlings, as seedling survival is largely enhanced after snowmelt, when germination is triggered by a relatively high temperature in the late spring or early summer (Alvarado and Bradford, 2002; Shimono and Kudo, 2005; Zhang et al., 2020).

The optimum temperature for the germination of P. kansuensis from cool habitats was slightly higher than that of the seeds from warm habitats, but the ceiling temperature showed no difference among provenances. All test seeds germinated well at a temperature interval of 20–25°C, suggesting that a low temperature during seed maturation and dispersal could prevent the emergence of seedlings and their death by freezing the following winter (Fenner and Thompson, 2005; Wesche et al., 2006). However, this study was inconsistent with other studies (Wang et al., 2008; Sui et al., 2013; Hu et al., 2022) in terms of the optimum temperature for P. kansuensis germination. Hu et al. (2022) showed that P. kansuensis germination was significantly higher at 25°C than at 20°C. The variation in the optimum temperature for germination in the studied populations of P. kansuensis may be related to genetic differentiation, environmental conditions, or the interactions of the two. The maternal plant’s genetics and environmental conditions during seed maturation and dispersal are considered crucial factors in the difference in germination requirements among the populations of a single species (Fenner and Thompson, 2005; Baskin and Baskin, 2014). Previous studies have found that annual temperature is one of the most important environmental variables determining the distribution of P. kansuensis (Sui et al., 2013; Wang et al., 2019), suggesting that the habitat occupied by this weed will continue to grow in western China due to rising temperatures (Wang et al., 2019).

Seeds adapted to germinating under water stress conditions are generally considered to have an advantage in germinating in arid or semiarid environments (Fenner and Thompson, 2005). However, there may be a trade-off between seed germination under low water potential and seedling survival after germination; seeds with higher germination at reduced water potential often have relatively low seedling establishment in arid regions, as seedlings require more water to compensate for evaporation (Flores and Briones, 2001). Previous studies have suggested that the relationship between seed germination and water stress/tolerance is largely species-specific and dependent upon environmental conditions (Fenner and Thompson, 2005; Baskin and Baskin, 2014). For example, Evans and Etherington (1990) reported that species originating from wetlands could not germinate well at low water potentials. By contrast, Rumex crispus from arid habitats could germinate when the water potential level was down to –1.5 MPa (Evans and Etherington, 1990). Zeng et al. (2010) found that species from semiarid regions achieved relatively higher germination levels at −1.8 MPa than those from arid habitats. Therefore, the seeds of species from humid habitats should be more sensitive to low water potentials than those from arid habitats, because the selection pressure from soil stress for seeds originating form humid habitats is lower than that of seeds originating from arid habitats (Ludewig et al., 2014).

Different populations of the same species may demonstrate different germination abilities when they germinate under water stress (Baskin and Baskin, 2014; Lima and Meiado, 2017). Previous studies found that the seed germination of Pilosocereus catingicola, Betula pendulai, and Pinus brutia from dry habitats was higher than that of the same plants from wet habitats with low water potentials (Lima and Meiado, 2017; Tilki and Dirik, 2007; Ranno et al., 2020). Similarly, Zhang et al. (2017) reported that the germination of Stipa bungeana from Western China’s Loess Plateau decreased with decreased water potential in eight populations, and that sensitivity to water stress varied among the populations. In the present study, the germination of P. kansuensis from different populations also differed when the seeds were subjected to water stress. The populations’ sensitivity to water stress is ranked as follows: HB (−0.68 MPa)< LZ (−0.84 MPa)< GL (−0.90 MPa)< SN (−1.03 MPa)< YS (−1.31 MPa), this ranking indicates that habitat had no plasticity to seed germination of P. kansuensis response to water stress. This finding is consistent with Daws et al. (2008), who detected no clear pattern in sensitivity to water potential in relation to habitat type for the germination of Neotropical species. The possible reasons for the within-population differences of P. kansuensis germination under water stress may be related to genetic differentiation and the moisture of the soil in which the maternal plants produced their seeds (Li et al., 2016). Recent studies have indicated that gene flow among P. kansuensis populations is limited to the Qinghai–Tibet Plateau (Li et al., 2016). Consequently, a high level of genetic differentiation has been shown among the populations (Li et al., 2019). Furthermore, the distance between collection sites was greater than 450 km, and the five populations likely had genetic differences. Therefore, genetic differences may explain, in part, the population differences in seed germination under low water potential (Zhang et al., 2017). Variations in soil water across seed collecting sites can also explain the differences in germination among the populations (Wang et al., 2019; Yang et al., 2020), as the annual precipitation of each habitat is significantly different. However, seed germination may be determined by other features of the maternal environment (Fenner and Thompson, 2005; Baskin and Baskin, 2014). Therefore, other environmental factors should also be considered when studying the germination characteristics of a single species across a wide geographical range.

A negative relationship has been noted between θ _H and Ψ _b(50) at a given temperature when testing the germination response at different water potentials. This suggests that seeds with a low θ _H and high Ψ _b(50) may germinate quickly under sufficient water conditions but prevented from germination under water stress conditions (Bradford, 1999; Batlla and Benech-Arnold, 2004; Zhang et al., 2020; Ullah et al., 2022). The present study supports these findings: the seeds originating from YS and SN, with high tolerance to water stress (low Ψ _b(50)), spent more time germinating than those from HB and LZ, with high Ψ _b(50). This gives seeds with a high θ _H on the Qinghai–Tibet Plateau an ecological advantage because seeds germinate during the driest period of the growing season (May) in this region (Hu et al., 2015). A high θ _H would inhibit germination after a rainfall event and prevent the death of seedlings after a subsequent drought (Zhang et al., 2020).