Introduction
Seed germination is highly regulated through environmental cueing (Donohue et al. Reference Donohue, Rubio de Casas, Burghardt, Kovach and Willis2010), and weeds often exhibit high levels of variation in timing and extent of germination (Baskin and Baskin Reference Baskin and Baskin1985; Egley and Williams Reference Egley and William1991; Forcella et al. Reference Forcella, Arnold, Sanchez and Ghersa2000). Variation in seed germination is determined, however, not only by the immediate microenvironment of the seed but also by the conditions experienced by parental plants (mainly maternal; Baskin and Baskin Reference Baskin and Baskin2019) during the preceding growing season (Fenner Reference Fenner1991; Roach and Wulff Reference Roach and Wulff1987). Maternal environment effects on seed germination have been reported for a wide range of plant species in response to temperature (Donohue et al. Reference Donohue, Heschel, Butler, Barua, Sharrock, Whitelam and Chiang2008), soil moisture (Wright et al. Reference Wright, Seavers, Peters and Marshall1999), CO2 (Farnsworth and Bazzaz Reference Farnsworth and Bazzaz1995), daylength (Nurse et al. Reference Nurse, DiTommaso and Ramirez2004), solar irradiance (Adkins and Armstrong Reference Adkins and Armstrong2007), light quality (Van Hinsberg Reference Van Hinsberg1998), nutrients (Allison Reference Allison2002), herbicides (Andersson Reference Andersson1996), competition (Nurse and DiTommaso Reference Nurse and DiTommaso2005), viral infection (Kathiria et al. Reference Kathiria, Sidler, Golubov, Kalischuk, Kawchuk and Kovalchuk2010), and even presence (or absence) of earthworms in the soil (Laossi et al. Reference Laossi, Noguera and Barot2010).
Variability in germination (dormancy) patterns acts as a bet-hedging strategy whereby some weed cohorts can escape control measures or unsuitable environments. Repeated applications (e.g., of herbicides) during the season are often necessary to control the staggered cohorts that result from temporally nonuniform germination patterns. Therefore, understanding what regulates the observed variation in germination patterns not only is a fundamental question in weed seed ecology but also has implications for management of weeds.
Environmental maternal effect adds another layer of complexity to germination studies, as the observed responses to current experimental factors are not independent of seed’s past exposure to different environmental conditions. Maternal effect, therefore, markedly constrains our ability to predict weed seed germination (and seedling emergence), as current germination/emergence models are developed solely using the present microenvironment experienced by the seed. The inconsistencies that we sometimes observe in germination studies are often interpreted as experimental errors but in part may have sound biological underpinnings due to maternal effects.
Interestingly, it seems that seed germination responds to certain maternal environments in a predictable manner. A literature review by Fenner (Reference Fenner1991) showed that high temperatures, short days, red light, drought, and high nitrogen during seed development were consistently associated with increased germinability in the progeny (Fenner Reference Fenner1991). A good example of such predictable changes is an early study by Von Abrams and Hand (Reference Von Abrams and Hand1956) on hybrid roses that found a strong (positive) linear relationship (R2 ≈ 0.99) between the progeny seed germination and the mean average daily temperature 30 d before seed harvest. These observations suggest that the predictive power of germination (emergence) models can be improved by incorporating the information about the environment of the previous growing season. Such “transgenerational models” can be developed by quantifying changes in the attributes of an established model as a function of certain maternal environmental variables (e.g., soil moisture levels in which the mother plants have grown).
In this study, we examined the effects of maternal water environments on properties of the hydrotime model of germination in progeny seeds of Palmer amaranth (Amaranthus palmeri S. Watson). Hydrotime models have been successfully used to describe the temporal pattern of germination in response to reduced water in many species, including weeds (Batlla and Benech-Arnold Reference Batlla and Benech-Arnold2004; Bradford Reference Bradford1990; Bradford and Still Reference Bradford and Still2004; Gianinetti and Cohn Reference Gianinetti and Cohn2007; Gummerson Reference Gummerson1986). Specifically, our objective was to quantify potential changes in base water potential, Ψb(g), for seed germination of A. palmeri in response to maternal water stress. Base water potential represents a water potential threshold at or below which no germination will occur. As a key parameter in the hydrotime model, Ψb(g) measures the germination tolerance of species (population) to drought: lower (more negative) Ψb(g) values indicate that seeds can germinate from drier conditions (e.g., soils). The present study tests the hypothesis that plants grown under water stress will produce seeds characterized with lower Ψb(g) values and will therefore be capable of germinating under drier conditions compared with seeds originating from well-watered plants. This study represents the first step in developing transgenerational germination models that incorporate environmental conditions before and after seed dispersal. This study used A. palmeri, which is one of the most troublesome weeds in the United States, and hopefully could help with the development of decision support systems for effective management of this pervasive weed.
Materials and Methods
Seeds of two populations of A. palmeri, one from California (CA) and one from Kansas (KS) were provided by Anil Shrestha (California State University, Fresno, CA) and Dallas E. Peterson (Kansas State University, Manhattan, KS), respectively. A greenhouse experiment was conducted at the Orchard Parks greenhouse facilities of the University of California, Davis, during the summer of 2018. Round plastic pots (2.37 L) were filled with a soil mix (1:1 sand/peat) plus a controlled-release fertilizer (15–9–12, 150 g 75 L−1; Scotts Osmocote® PLUS, Mississauga, ON, Canada). Approximately 10 seeds from each A. palmeri population were sown into a total of 200 pots, 100 for each population. Plants were grown in a greenhouse with a temperature range of 32/22 C (day/night) and daylength of 16 h provided through supplementary lighting. Seedlings were thinned randomly to achieve one plant per pot by the 4-leaf stage. All plants received normal irrigation using four emitters inserted into the potting medium to deliver 65 ml of water min−1 for 2 min, twice per day (7:00 AM and 2:00 PM). For each population, when plants reached the 6-leaf stage, 50 pots (replicates) were randomly assigned to water-stress treatment by removing three out of four drip emitters, which reduced the amount of water by ~75% (field capacity 6% to 13%). The remaining 50 pots continued to receive normal irrigation (field capacity 35% to 42%), as described earlier. Seeds were harvested at full maturity as determined by inspecting the inflorescences for the presence of black, firm seeds (Mesgaran et al. Reference Mesgaran, Matzrafi and Ohadi2019).
Seed Weight and Size Measurements
One month after harvest, 10 seed subsamples from each population by maternal water treatment combinations, containing 100 seeds each, were weighed using a precision scale (ME104E™, Mettler-Toledo, Columbus, OH) to obtain 1,000-seed weight data. Seeds (40 to 70 seeds) were imaged using a Leica M205 FA stereomicroscope (Leica Microsystems, Wetzler, Germany) equipped with a digital camera (Leica MC190 HD Camera) and built-in image analytical software (LAS X Core) to measure the seed size based on the surface area.
Seed Germination Tests
Germination (both parental and progeny generations) was tested under five nominal water potentials (0, −0.2, −0.4, −0.6, and −0.8 MPa) at two constant temperatures of 20 and 30 C representing suboptimal and optimal temperatures for germination of A. palmeri, respectively (Guo and Al-Khatib Reference Guo and Al-Khatib2003; Steckel et al. Reference Steckel, Sprague, Stolle and Wax2004). Experiments were conducted with four replicates (petri dishes) per treatment and repeated twice. Experimental runs were rotated between two growth chambers (Conviron® CMP 6010, Winnipeg, Canada). Solutions were prepared by dissolving appropriate amounts of polyethylene glycol 8000 (Sigma-Aldrich, St. Louis, MO) in deionized water according to Michel (Reference Michel1983). Filter papers (Whatman® Grade 2; Sigma-Aldrich) were soaked in each solution containing the desired water potential for 12 h before the experiment, as recommended by Hardegree and Emmerich (Reference Hardegree and Emmerich1990). The water potential of each solution was tested twice: before and after adding the filter papers to the solution using a Wescor Vapro osmometer (Wescor, Logan, UT, USA). For each treatment, 40 seeds from each population were placed in a 9-cm petri dish lined with moistened filter paper and sealed with Parafilm® to reduce water loss from the dishes.
Hydrotime Model and Statistical Analysis
The hydrotime model is based on a linear relationship between germination rate, GR (i.e., the inverse of time to a given germination percentile, 1/t g) and the medium water potential, Ψ (Bradford Reference Bradford2002; Gummerson Reference Gummerson1986):
$${{{\rm{GR}} = }\frac{{1}\over{{{t_g}}}} {= \frac{{{\left( {{\rm{\Psi }} - {{\rm{\Psi }}_{b\left( g \right)}}} \right)}}\over{{{{\rm\theta }}_H}}}{\rm{\;\;\;\;\;}}}}$$
where tg is time to germination of a given percentile g (the subscript “g” indicates the germination percentile), θH is hydrotime constant describing the amount of time (in MPa day or MPa hour units) required for germination of any fractions of the seed subpopulation. The main feature of the hydrotime model is that base water potential varies among seeds (hence the subscript “g” in Ψb(g)), giving rise to varaiton in timing of germination for each individual seed, as shown below after rearranging Equation 1 for Ψb(g):
$${{\rm{\Psi }}_{b\left( g \right)}} = {\rm{\Psi }} - \left( {\frac{{{{\rm{\theta }}_H}}}\over{{{t_g}}}} \right)$$
Variability in Ψb(g) in a seed population can be described by various statistical frequency distributions (Mesgaran et al. Reference Mesgaran, Mashhadi, Alizadeh, Hunt, Young and Cousens2013; Watt et al. Reference Watt, Xu and Bloomberg2010), but most often is assumed to be normally distributed with its respective mean (median), Ψb(50), and standard deviation, σΨb (Bradford Reference Bradford1990, Reference Bradford2002). Assuming a normal distribution, the cumulative germination of seeds, g, over time, t, at any given water potential, Ψ, can be obtained by:
$$g\left( {{\rm{\Psi }}, {t_g}} \right) = \;{\rm{\Phi }}\left( {{{\rm{\Psi }}_{b\left( g \right)}}, {{\rm{\Psi }}_{b\left( {50} \right)}}, {{\rm{\sigma }}_{{\rm{\Psi }}b}}} \right)\;\;\;\;\;$$
Replacing Ψb(g) with Equation 2 gives:
$$g\left( {{\rm{\Psi }}, {t_g}} \right) = {\rm{\;\Phi }}\left( {{\rm{\Psi }} - \left( {\frac{{{{\rm{\theta }}_H}}}\over{{{t_g}}}} \right), {{\rm{\Psi }}_{b\left( {50} \right)}}, {{\rm{\sigma }}_{{\rm{\Psi }}b}}} \right){\rm{\;\;\;\;\;}}$$
where Φ is the cumulative distribution function of a normal distribution. Equation 3 was fit to germination data using the drcSeedGerm package (Onofri et al. Reference Onofri, Benincasa, Mesgaran and Ritz2018) in R (R Core Team 2020). The hydrotime model (Equation 3) was fit to germination data of each temperature and population separately. To obtain robust standard errors for model parameters, we implemented the delete-a-group jackknife technique (Onofri et al. Reference Onofri, Mesgaran, Neve and Cousens2014) using the jackGroupSE() function available in drcSeedGerm package (Onofri et al. Reference Onofri, Benincasa, Mesgaran and Ritz2018). This method provides more accurate estimates of standard errors for cumulative data such as germination curves (Onofri et al. Reference Onofri, Mesgaran, Neve and Cousens2014). To test whether the germination curves differed between the two maternal watering levels, a reduced hydrotime model that ignored the maternal treatment as grouping factor was fit to the data. Using the anova() function in R (R Core Team 2020), an F-test was conducted to test whether the size of error in the reduced model is significantly larger than that of the full model: the full model included the maternal water environment as a grouping factor. A significant P-value (≤0.05) indicates that the germination curves significantly differ between the two maternal water conditions.
Parental plant height and progeny seed weight and seed size were compared across the two maternal water treatments (water-deficit vs. well-watered) using the t-test with Proc TTEST in SAS (v. 9.4; SAS Institute 2019). As a surrogate measure of dormancy level, the total percent germination at 0 MPa and 30 C treatment, representing the optimal conditions for germination of A. palmeri (Steckel et al. Reference Steckel, Sprague, Stolle and Wax2004), was further analyzed using generalized linear mixed models with PROC GLIMMIX in SAS (v. 9.4). In this analysis, a binomial distribution with the logit function was used, and maternal water treatment was considered the fixed effect, while the experimental run was incorporated as the random effect. Data from each population were analyzed separately. Visual assessment of residuals (plotted against fitted values) showed no evidence of variance heteroscedasticity.
Results and Discussion
Our irrigation scheme was effective in causing stress in the water deficit treatment, as evidenced by significant reduction in plant height of both A. palmeri populations (P-value < 0.0001). Plant height was almost halved in plants exposed to the water deficit treatment (Figure 1).
Figure 1. Plant height for Amaranthus palmeri populations (California and Kansas) grown under continuous water-deficit or well-watered irrigation conditions. Vertical lines on bars indicate SE.
Seed Weight and Area
Analysis of progeny 1,000-seed weight showed significant differences between the two maternal water treatments both in the California (P-value = 0.0003) and the Kansas (P-value = 0.0003) populations. In both populations of A. palmeri, seeds produced under water stress were 18% heavier than those produced by plants under normal irrigation (Figure 2). The 1,000-seed weights of original (parental) seeds were 290.7 mg (SE: 8.5) and 279.5 mg (SE: 8.2) for the California and Kansas collections, respectively.
Figure 2. Differences in 1,000-seed weight, seed surface area, and total germination (dormancy) of progeny seeds from two Amaranthus palmeri populations (California and Kansas) grown under continuous water-deficit (WD) or well-watered (WW) irrigation conditions. Vertical bars on data points indicate SE.
Progeny seed size, as measured by the seed surface area, differed between maternal water treatments in the California population (P-value = 0.0117) but not in the Kansas population (P-value = 0.7534). In the California population, progeny seed surface area increased by 23% in response to maternal water stress (Figure 2). The surface areas of parental seeds used in the study were 0.656 mm2 (SE: 0.023) for the California population and 0.621 mm2 (SE: 0.037) for the Kansas population. As seed surface area in the Kansas population was constant across the two maternal conditions, the increase in seed mass observed with the water-deficit treatment should have resulted from the increase in seed thickness and/or density. Previous studies with crop plants such as lentil (Lens culinaris Medik.) (Sehgal et al. Reference Sehgal, Sita, Kumar, Kumar, Singh, Siddique and Nayyar2017), soybean [Glycine max (L.) Merr.] (Wijewardana et al. Reference Wijewardana, Raja Reddy, Jason Krutz, Gao and Bellaloui2019), and canola (Brassica napus L.) (Hatzig et al. Reference Hatzig, Nuppenau, Snowdon and Schießl2018) have shown that applying water stress to parental plants may negatively affect yield-related traits and reduce final seed weight. In cereal crops such as rice (Oryza sativa L.), maize (Zea mays L.), and wheat (Triticum aestivum L.), water stress at the grain-filling stage translated to a shorter grain-filling period, eventually leading to lower seed weight (Saini and Westgate Reference Saini and Westgate1999). However, similar to our findings, a greenhouse study with redroot pigweed (Amaranthus retroflexus L.) showed that plants subjected to water stress produced heavier seeds (greater mass) than those grown without drought (Chadoeuf-Hannel and Barralis Reference Chadoeuf-Hannel and Barralis1982). Greater seed mass in A. retroflexus was also associated with drier environments in the correlative study of Schimpf (Reference Schimpf1977). Three of four Polygonum spp. (i.e., ladysthumb [Polygonum persicaria L.], tufted knotweed [Polygonum cespitosum Blume], and pale smartweed [Polygonum lapathifolium L.]) tested by Sultan (Reference Sultan2001) produced larger achenes in dry rather than moist environments. Interestingly, the noninvasive species marshpepper smartweed (Polygonum hydropiper L.) was the only species that produced smaller achenes in dry soils. Whereas our study shows that maternal drought can markedly affect seed mass and size in A. palmeri, shading, even as high as 87%, had no effect on seed weight of A. palmeri (Jha et al. Reference Jha, Norsworthy, Riley and Bridges2010).
Seed Dormancy
Progeny seed dormancy was evaluated by testing the germination of A. palmeri seeds at optimal temperature (i.e., 30 C; Steckel et al. Reference Steckel, Sprague, Stolle and Wax2004). Maternal water stress significantly affected the dormancy level in progeny seeds of both the California (P-value = 0.0004) and Kansas (P-value < 0.0001) populations. Plants that experienced water stress produced the least dormant seeds (Figure 2). For example, in the Kansas population, the total germination for seeds originating from water-stressed plants was as high as 87.5% (SE: 2.6) but dropped to 55.6% (SE: 3.7) in seeds from plants receiving normal irrigation (Figure 2). In the California population, the total germination for seeds from the water-deficit treatment was 20% higher than for the control. The parental seeds used in the study exhibited dormancy levels similar to those produced by well-watered plants: California parental seeds germinated 30.6% (SE: 2.3), while parental seed sourced from Kansas reached 65.0% (SE: 3.7) germination.
In several other weed species such as johnsongrass [Sorghum halepense (L.) Pers.] (Arnold et al. Reference Arnold, Fenner and Edwards1992) and A. retroflexus (Karimmojeni et al. Reference Karimmojeni, Bazrafshan, Majidi, Torabian and Rashidi2014), when parental plants were exposed to water stress during seed production, progeny seeds showed increased germination rates compared with seeds of plants grown under control irrigation conditions. However, an opposite trend has been recorded for germination of seeds from maternally stressed plants in crops. In soybean, increasing maternal drought stress translated into lower seed germination rates (Wijewardana et al. Reference Wijewardana, Raja Reddy, Jason Krutz, Gao and Bellaloui2019). For canola, differences in mean germination time between seeds of genotypes grown under water stress was not conclusive, although some genotypes grown under water stress took longer to germinate than the control (Hatzig et al. Reference Hatzig, Nuppenau, Snowdon and Schießl2018). Differences between crop plants and weedy species may be explained due to evolutionary changes related to historical domestication and breeding processes. One of the main features of plants that have undergone strong domestication processes is reduced seed dormancy through changes in seed coat thickness (e.g., physical dormancy) or biochemical seed composition (e.g., physiological dormancy) (Abbo et al. Reference Abbo, Pinhasi van-Oss, Gopher, Saranga, Ofner and Peleg2014). Different drought stress responses were reported when weedy and domesticated populations of sunflower (Helianthus annuus L.) were compared (Koziol et al. Reference Koziol, Rieseberg, Kane and Bever2012).
Seed germination is regulated by several plant hormones, mainly abscisic acid (ABA) and gibberellic acid (GA) accumulated in the embryo during seed development (Bewley Reference Bewley1997; Kermode Reference Kermode2005). Physiological seed dormancy is regulated by ABA/GA ratio, where ABA prevents and GA promotes seed germination. ABA level increases during initial seed development and decreases during seed maturity; however, this process can be affected by environmental signals (Bewley Reference Bewley1997). In grain sorghum [Sorghum bicolor (L.) Moench.] seeds, water stress decreased both embryo content and sensitivity to ABA (Benech Arnold et al. Reference Benech Arnold, Fenner and Edwards1991). Elevated ABA levels were recorded in seeds of A. palmeri plants grown under 87% shade compared with 0% shade conditions (Jha et al. Reference Jha, Norsworthy, Riley and Bridges2010).
Hydrotime Germination Model
With few exceptions, the hydrotime model adequately described the germination time courses of A. palmeri at both tested temperatures as shown by fitted curves (Figures 3 and 4) and small root mean-square error (RMSE) as shown in Table 1. There were marked differences in progress toward germination between the two maternal water conditions. In both A. palmeri populations, seeds from water-stressed plants achieved greater germination percentages (almost double) than those from plants with no water stress across all levels of water potentials tested in this study (Figures 3 and 4).
Figure 3. The effect of maternal water conditions (continuous water-deficit vs. well-watered) on cumulative germination of progeny seeds in two Amaranthus palmeri populations (California and Kansas) tested under various water potentials at 20 C. Lines are fitted values obtained from the hydrotime model, Equation 3, 
$$\;g\left( {{\rm{\Psi }}, {t_g}} \right) = {{\it\Phi} \left( {{\rm{\Psi }} - \left( {\frac{{{{\rm{\theta }}_H}}}\over{{{t_g}}}} \right), \ {{\rm{\Psi }}_{b\left( {50} \right)}}, \ {{\rm{\sigma }}_{{\rm\Psi }}b}}} \right)$$
, with parameter estimates shown in Table 1.
Figure 4. The effect of maternal water conditions (continuous water-deficit vs. well-watered) on cumulative germination of progeny seeds in two Amaranthus palmeri populations (California and Kansas) tested under various water potentials at 30 C. Lines are fitted values obtained from the hydrotime model, Equation 3, 
$$g\left( {{\rm{\Psi }}, {t_g}} \right) = \it\Phi \left( {{\rm{\Psi }} - \left( {\frac{{{{\rm{\theta }}_H}}}\over{{{t_g}}}} \right), {{\rm{\Psi }}_{b\left( {50} \right)}}, {{\rm{\sigma }}_{{\rm{\Psi }}b}}} \right), \ $$ with parameter estimates shown in Table 1.
Table 1. Parameter estimates of hydrotime model of germination, Equation 3,
$$\;g\left( {{\rm\Psi }}, {t_g}} \right) = \;{\rm\Phi} \left( {{\rm{\Psi }} - \left( {\frac{{{{\rm{\theta }}_H}}}\over{{{t_g}}}} \right), {{\rm{\Psi }}_{b\left( {50} \right)}}, {{\rm{\sigma }}_{{\rm{\Psi }}b}}} \right)$$, for progeny seeds of two Amaranthus palmeri populations (California and Kansas) grown under contrasting maternal water conditions.a
a Germination assays were conducted at two temperatures (20 and 30 C), and the hydrotime model was fit to data of each temperature separately. Also shown are the parameter estimates for the original (parental) seed sources used to grow plants under the two maternal water treatments. Values in the parenthesis indicate SE.
b RMSE, root mean-square error. Note that a single hydrotime model that included the maternal water environment as a covariate was fit to the germination data of each population. As result, there is one RMSE for both maternal water conditions.
c P-value indicates the overall significance level for the difference between the germination curves (models) of the two maternal water levels, i.e., water-deficit vs. well-watered treatments.
The observed changes in germination capacity of seeds (Figures 3 and 4) can largely be attributed to shifts in population base water potential (Ψb(50)) and to a lesser extent to its standard deviation (i.e., σΨb), as hydrotime constant (θH) was not affected by the maternal environment (Table 1). Base water potential exhibited a consistent pattern of response to maternal water treatments across the two tested temperatures and in both A. palmeri populations. Maternal drought significantly shifted the distribution of base water potential in progeny seed subpopulations toward more negative values (Figure 5), implying that the resultant seeds have become more tolerant to water limitation and hence can germinate from drier conditions. For example, in the Kansas population, the estimated Ψb(50) was −0.63 MPa (SE: 0.036) at 30 C when seeds matured on water-stressed plants, which was three times more negative than the Ψb(50) (i.e., −0.21 MPa, SE: 0.025) of seeds produced by nonstressed plants. These Ψb(50) values suggest that if seeds from nonstressed plants are planted into a medium with a water potential of −0.63 MPa, only 18% of them are predicted to germinate as contrasted to 50% germination for seeds from water-stressed plants. Maternal water stress also reduced the variance of Ψb(g) in seed subpopulations as indicated by smaller σΨb values with the only exception being the progeny seed from California population tested at 30 C (Table 1; Figure 5). As parameter σΨb in the hydrotime model is a measure of germination uniformity, reduction in σΨb implies a higher uniformity for A. palmeri seeds originating from maternally stressed plants.
Figure 5. Shifts in distribution of base water potential of progeny seeds from two Amaranthus palmeri populations (California and Kansas) grown under contrasting maternal water conditions (continuous water-deficit vs. well-watered). Seed germination was tested at two temperatures (20 and 30 C) under five water potentials. A hydrotime model, Equation 3, 
$$g\left( {{\rm{\Psi }}, {t_g}} \right) = \it\Phi \left( {{\rm{\Psi }} - \left( {\frac{{{{\rm{\theta }}_H}}}\over{{{t_g}}}} \right) {{\rm{\Psi }}_{b\left( {50} \right)}}, {{\rm{\sigma }}_{{\rm{\Psi }}b}}} \right), \ $$was fit to estimate the median base water potentials, Ψb(50) (vertical dashed lines), and their respective standard deviations, σΨb, to produce these probability density curves of normal distribution (see Table 1 for parameter estimates). Note that the area under the curve for base water potential values >0 indicates the proportion of seeds that have not germinated (i.e., dormancy level).
We are only aware of a single study that has examined the effect of maternal water level on the germination tolerance of progeny seed to water limitation. Seed of wild tomato (Lycopersicon chilense Dunal) (an endemic tomato species in Chile) from mother plants grown under water deficit were able to germinate at a lower water potential than those from plants receiving ample water supply (Maldonado et al. Reference Maldonado, Pujado and Squeo2002). Low temperature during seed maturation reduced the ability of mouseear cress [Arabidopsis thaliana (L.) Heynh.] to germinate at low water potential (Edwards et al. Reference Edwards, Burghardt, Zapata-Garcia and Donohue2016).
Our results suggest that a transgenerational hydrotime model can be developed by adjusting Ψb(50), and perhaps σΨb, to adequately describe the progeny seed germination. A shift in Ψb(50) toward more negative values has been associated with reduced dormancy in other studies (Batlla and Benech-Arnold Reference Batlla and Benech-Arnold2004; Bradford Reference Bradford2002). For example, temporal change in seed dormancy of prostrate knotweed (Polygonum aviculare L.) was also modeled by using Ψb(50) alone, as the θH and σΨb parameters changed only slightly with dormancy level (Batlla and Benech-Arnold Reference Batlla and Benech-Arnold2004).
Future studies need to use multiple levels of maternal water limitation to determine the shape of relationships between maternal water level and Ψb(50) and σΨb; for example, whether they are linear or nonlinear. Our study used continuous water stress (from 6-leaf stage onward); however, it is likely that only stress at the reproductive stage (i.e., when seed develops) will be important in driving maternal effects, as shown in several studies (reviewed by Baskin and Baskin Reference Baskin and Baskin2014). Applying water stress at different parental growth stages can inform us about the most influential stage that can produce maternal effects.
Conclusions
Our results show that A. palmeri plants experiencing water stress during their growth can produce less dormant and larger seeds that may give rise to competitive and vigorous seedlings. Moreover, we showed, for the first time, that A. palmeri parental plants grown under water-deficit conditions can produce seeds that are more drought tolerant and hence can germinate under drier conditions compared with seeds from nonstressed plants. These maternal effects seem to be adaptive, as the mother plant exposed to stress produced progeny somewhat primed (prepared) to cope with the harsh environments similar to maternal conditions. The consistency in germination responses of progeny seed to maternal conditions suggests that further models of germination (and emergence) need to be transgenerational, as the “memory” of past environment can have a marked influence on the responses of seeds to current conditions.
Acknowledgments
We wish to thank Anil Shrestha (California State University, Fresno, CA) and Dallas E. Peterson (Kansas State University, Manhattan, KS) for providing the seed used in the greenhouse experiments. The authors would also like to thank Hannah Clifton and William Werner for greenhouse maintenance. This research was funded by a UC Davis New Research Initiatives and Interdisciplinary Research Grant. No conflicts of interest have been declared.
Open access
