PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of annbotAboutAuthor GuidelinesEditorial BoardAnnals of Botany
 
Ann Bot. 2009 June; 103(8): 1291–1301.
Published online 2009 March 29. doi:  10.1093/aob/mcp070
PMCID: PMC2685320

Germination parameterization and development of an after-ripening thermal-time model for primary dormancy release of Lithospermum arvense seeds

Abstract

Background and Aims

Models based on thermal-time approaches have been a useful tool for characterizing and predicting seed germination and dormancy release in relation to time and temperature. The aims of the present work were to evaluate the relative accuracy of different thermal-time approaches for the description of germination in Lithospermum arvense and to develop an after-ripening thermal-time model for predicting seed dormancy release.

Methods

Seeds were dry-stored at constant temperatures of 5, 15 or 24 °C for up to 210 d. After different storage periods, batches of 50 seeds were incubated at eight constant temperature regimes of 5, 8, 10, 13, 15, 17, 20 or 25 °C. Experimentally obtained cumulative-germination curves were analysed using a non-linear regression procedure to obtain optimal population thermal parameters for L. arvense. Changes in these parameters were described as a function of after-ripening thermal-time and storage temperature.

Key Results

The most accurate approach for simulating the thermal-germination response of L. arvense was achieved by assuming a normal distribution of both base and maximum germination temperatures. The results contradict the widely accepted assumption of a single Tb value for the entire seed population. The after-ripening process was characterized by a progressive increase in the mean maximum germination temperature and a reduction in the thermal-time requirements for germination at sub-optimal temperatures.

Conclusions

The after-ripening thermal-time model developed here gave an acceptable description of the observed field emergence patterns, thus indicating its usefulness as a predictive tool to enhance weed management tactics.

Key words: Lithospermum arvense, winter annual weed, thermal-time model, cardinal temperatures, primary dormancy, after-ripening thermal-time, storage temperature, field germination, seedling emergence

INTRODUCTION

Lithospermum arvense L. [= Buglossoides arvensis (L.) I.M.Johnst.] is a dicotyledonous facultative winter annual plant, native to Eurasia. In Argentina, particularly in the semi-arid area of the Buenos Aires Province, L. arvense has become an increasingly important weed in winter crops, mainly due to the adoption of low-impact tillage systems during the last few years (Chantre et al., 2009).

At seed maturity, L. arvense seeds (nutlets) show physiological dormancy, requiring an after-ripening period for dormancy release (Baskin and Baskin, 1998). According to Baskin and Baskin (1976, 1986, 1998), after-ripening is the process by which winter annual species lose primary dormancy after exposure of dry seeds to warm temperatures. Temperature is the major environmental factor governing changes in the degree of dormancy in temperate environments (Benech-Arnold et al., 2000); in general, the higher the storage temperature, the greater the loss of dormancy (Thompson, 1973; Bell, 1993; Murdoch and Ellis, 2000). Allen et al. (1995) and Bauer et al. (1998) found that after-ripening increased both the percentage and the germination rate of Bromus tectorum seeds. Favier (1995) proposed a germination rate model for dormancy loss of Hordeum vulgare seeds where the rate of change of mean germination time was described as function of after-ripening temperature. Using a thermal-time approach, Steadman et al. (2003a, b) found that the rate of dormancy loss in Lolium rigidum seeds was a positive linear function of after-ripening temperature.

Many thermal-germination models have been constructed on the basis of seed-to-seed response variation in order to understand the behaviour of the seed population. Most thermal-germination models use cardinal temperatures (base, optimal and maximal) and sub-optimal (θ1) and supra-optimal (θ2) thermal-times to quantify within-population variability (Garcia-Huidobro et al., 1982; Covell et al., 1986; Ellis et al., 1986, 1987; Ellis and Butcher, 1988; Murdoch et al., 1989; Bradford, 1996; Hardegree et al., 1999). The most accepted cardinal thermal-germination approach assumes a constant value of base temperature (Tb) for the entire seed population, and a normal or log-normal distribution of θ1 in the sub-optimal temperature range (Covell et al., 1986; Ellis et al., 1986; Benech-Arnold et al., 1990; Steinmaus et al., 2000; Alvarado and Bradford, 2002). In the supra-optimal thermal range, a common value of θ2 and a normal or log-normal distribution of the maximal germination temperature (Tc) are usually accepted (Covell et al., 1986; Ellis et al., 1986; Ellis and Butcher, 1988). However, Hardegree (2006) obtained a better model fit using four rangeland grass species by assuming a common value of Tc and a log-normal distribution of θ2. Working with Stellaria media seeds, Grundy et al. (2000) proposed the idea of threshold temperatures for germination as a complement to the cardinal temperatures concept. According to this approach, each seed has lower and upper temperature thresholds for germination that are normally distributed in the population and define the temperature range within which each seed can germinate. Batlla and Benech-Arnold (2003) associated the concept of threshold temperatures to the dormancy status of a population of Polygonum aviculare seeds and successfully developed a thermal-time model for dormancy release.

The period of primary dormancy in seeds of winter annual species occurs during the summer season, and it coincides with the average length of the unfavourable season for seedling establishment (Schütz et al., 2002). As these seeds are released from dormancy, the permissive thermal-germination range widens and allows them to germinate at progressively higher temperatures (Karssen, 1982), such those prevailing during autumn after seed dispersal (Baskin and Baskin, 1986, 1998). Temperature is the main factor regulating seed dormancy status. As a result, the use of thermal-time models based on the accumulation of biological time can be very helpful to account for changes in the distribution of the population thermal parameters in relation to the dormancy-breaking conditions. The objectives of the present work were: (1) to evaluate the relative accuracy of different thermal-time models for the description of L. arvense germination under a range of constant temperatures, and (2) to develop an after-ripening thermal-time model for primary-dormancy release, based on the changes of the seed population thermal-time parameters.

MATERIALS AND METHODS

After-ripening treatments

Seeds of Lithospermum arvense L. [= Buglossoides arvensis (L.) I.M.Johnst.] were collected in a wheat field located in Bahía Blanca (38°44′s, 62°16′W), Argentina, at the time of their natural dispersal (December 2005). After harvest, seeds were cleaned and placed in paper bags before dry storage in growth chambers at constant temperatures of 5, 15 and 24 °C for up to 210 d. Seed germinability was assessed on freshly matured seeds and after 45, 100, 150 and 210d of dry storage. The water content of seeds at harvest was 8–10 % (d. wt basis).

Germination test

An aluminum temperature gradient bar designed according to Chatterton and Kadish (1969) was used for germination trials. Batches of 50 seeds were incubated at eight constant temperature regimes of 5, 8, 10, 13, 15, 17, 20 or 25 °C over a 30-d period. A 12-h photoperiod was applied by means of fluorescent lamps providing a photosynthetic photon flux density of 80 µmol m−2 s−1. Germination was monitored daily, visible protrusion of the radicle being the criterion for germination. Germination percentages were calculated for the viable fraction of the seed population. To test seed viability, ungerminated seeds were sliced longitudinally and incubated in 0·1 % (wt/vol.) tetrazolium chloride (2,3,5-triphenyltetrazolium chloride) solution for 24 h in the dark at 30 °C (International Seed Testing Association, 1999). Seeds with pink- or red-stained embryos were considered viable.

Determination of sub-optimal and supra-optimal temperature range for germination

Germination time-course curves obtained after different storage periods were used to estimate the time taken for cumulative germination to reach subpopulation percentiles of 10, 20, 30, 40, 50, 60, 70, 80 and 90 %. These estimates were obtained by linear interpolation between daily germination percentages at each incubation temperature. Germination rates at each percentile (reciprocals of time estimates of each fraction) were then plotted against incubation temperature to estimate the optimum temperature for germination (maximal germination rate) by means of a linear regression model. A linear relationship between germination rates and incubation temperature was assumed (Garcia-Huidobro et al., 1982; Covell et al., 1986). For the sub-optimal range,

equation image
1

while for the supra-optimal range,

equation image
2

where r is the germination rate (reciprocal of germination time of fraction g), T is the incubation temperature, Tb is the base temperature for germination, Tc is maximum or ‘ceiling’ temperature, and θ1 and θ2 are the sub- and supra-optimal thermal-time, respectively.

Optimum temperature (To) values were calculated for each subpopulation as the intercept of the sub- and supra-optimal temperature-response functions (Hardegree, 2006). Linear regression equations for each percentile were recalculated and constrained to pass through the average To value in order to identify the sub- and supra-optimal population thermal range for germination.

Parameterization of thermal-germination models

Experimentally obtained cumulative-germination curves were used to perform a non-linear regression procedure to assess the relative accuracy of different thermal-germination models in predicting germination response under constant incubation temperatures. Seed germination follows a binomial cumulative distribution function within the population due to seed-to-seed response variation. Thus, simulation of observed data was based on the cumulative distribution function of the normal distribution according to the ‘central limit theorem (Sokal and Rohlf, 1981).

Model 1

In this model a normal distribution of θ1 was assumed while Tb was considered constant for the entire population (Covell et al., 1986; Ellis et al., 1986, 1987). In the supra-optimal temperature range, Tc was considered to be normally distributed in the population while commonality of θ2 was assumed (Ellis et al., 1986). For the sub-optimal temperature range θ1 ~ N(θ50, σθ2), therefore the proportion of germinating seeds was described by the following equation:

equation image
3

where p is the proportion of germinating seeds at a given θ1 of a g fraction of the population, Φ is the normal probability integral, and θ50 and σθ are the mean and standard deviation of the normal distribution, respectively.

For the supra-optimal temperature range, germination prediction based on Tc ~ N(Tc50, σTc2) followed the equation:

equation image
4

where p is the proportion of germinating seeds at a given Tc of a g fraction of the population, Φ is the normal probability integral, and Tc50 and σTc are the mean and standard deviation of the normal distribution, respectively.

Model 2

A normal distribution of both Tb and Tc was assumed while θ1 and θ2 remained constant for all subpopulations. The equation for the sub-optimal temperature range is

equation image
5

where Tb50 and σTb are the mean and standard deviation of the normal distribution, respectively. For the supra-optimal range eqn (4) was applied.

Model 3

The germination-threshold approach proposed by Grundy et al. (2000) was used to estimate seed maximum germination percentages in relation to the after-ripening treatments. According to this approach each seed of the population is capable of germinating in a temperature range within a lower and an upper temperature threshold, and both are considered to be normally distributed in the population. The proportion of germinating seeds at a given temperature was calculated as:

equation image
6

where p(T) is the proportion of seeds germinating at temperature T. Means and standard deviations of the lower and upper threshold distributions were mL, SL and mu, Su, respectively.

Experimentally obtained germination curves were simulated within both temperature-threshold limits (germinable fraction of the population) assuming a normal distribution of both θ1 and θ2. Base and maximum germination temperatures were estimated from eqns (1) and (2).

Optimal thermal parameters for germination models were obtained by a non-linear least-squares curve-fitting method using an optimization program (Premium Solver Platform 7·0; Frontline systems, Inc). Maximum fit between simulated and experimentally obtained data was achieved by an iterative technique using a quasi-Newton algorithm.

Goodness-of-fit and statistical significance of thermal-germination models

Assessment of goodness-of-fit was performed by means of the coefficient of determination (R2). The fraction variance accounted for by the simulation model R2 was calculated as:

equation image
7

where yobs are observed values, ysim are simulated values. An R2 value of 1 indicates a perfect fit of the model to the observed data.

The relative accuracy of the models tested for the prediction of L. arvense germination response was evaluated in terms of the statistical significance of the comparison between R2 coefficients. The following F-test was used (Vohnout, 2003):

equation image
8

where R2 is the coefficient of the model that best fits the data, *R2 is the coefficient of the model to compare, n is the total number of cases in the model, p is the number of variables related to R2 and q is the number of variables related to *R2. The statistical significance was obtained from the F-distribution with q and n – p – 1 degrees of freedom.

Development of dormancy model

The process of model development is summarized below.

  1. Determine seed population thermal parameters that maximize the fit between the experimentally obtained germination time-course curves for seeds stored dry at 5, 15 and 24 °C during different after-ripening time periods and the predicted germination curves for each germination model.
  2. Assess the relative accuracy of each germination model in the prediction of the germination response by means of the determination of R2 coefficients and the statistical significance of their comparison.
  3. Select the germination model that best fits the observed data.
  4. Characterize changes in the seed population thermal parameters over time as a function of storage temperature, and derive thermal-time equations relating rate of change of these parameters to the after-ripening thermal-time.
  5. Using equations developed in (4), predict changes in the population thermal parameters of seeds after-ripened in the laboratory and in the field.
  6. Use predicted values of the population thermal parameters to simulate germination time-course curves and compare with experimentally obtained emergence data.

Model evaluation

In order to evaluate model performance with independent data, results from an experiment carried out under field conditions were used. Seeds of L. arvense were collected in a wheat field at Bahía Blanca, Argentina, at the time of their natural dispersal (January 2005). After harvest, seeds were cleaned and placed in paper bags before dry storage in growth chambers at constant temperatures of 20 and 40 °C. The water content of seeds at harvest was 6–8 % (d. wt basis). Lithospermum arvense seeds were divided into three different seed lots that were subjected to different after-ripening treatments. Seed lot 1 was stored at 20 °C for 33 d; seed lot 2 was stored at 20 °C for 20 d plus 30 d at 40 °C; and seed lot 3 was exposed to a constant temperature regime of 40 °C for 70 d. Subsequently, L. arvense seeds were buried at 1 cm depth in an experimental field of the CERZOS, Universidad Nacional del Sur and CONICET, located in Bahía Blanca, Argentina (38°39′54″s 62°13′58″W). Seeds belonging to each seed lot were buried in plots of 0·8 × 0·8 m following a completely randomized factorial design with three replicates (80 seeds per plot). Seed burial date varied according to the duration of each pre-burial after-ripening treatment; thus seeds were buried in the field on 24 February 2005 (seed lot 1), 11 March 2005 (seed lot 2) and 30 March 2005 (seed lot 3).

Seedling emergence was recorded every 2 weeks from early autumn until early spring in order to include the complete emergence season of this species. Seedlings were counted as emerged when cotyledons were visible, and seedlings were removed immediately after counting. Plots were irrigated periodically using a localized drip irrigation system in order to avoid effects on emergence caused by erratic precipitation in the region under study. Field temperature was recorded every 2 h at 1 cm depth using temperature data loggers (Thermochron Ibuttons, Model DS1921G-F50, Maxim Integrated Products, Inc).

RESULTS

Germination time-course curves and median germination rates

Maximum germination values of 50–60 % were observed for recently harvested seeds of L. arvense incubated at low temperatures (5–10 °C) while germination was almost negligible at incubation temperatures higher than 13 °C (Fig. 1A). As after-ripening time progressed, L. arvense seeds were able to germinate at higher incubation temperatures (compare rows in Fig. 1). The percentage of germinating seeds at a given time of the after-ripening process was clearly influenced by the storage temperature. Germination percentages were higher and seeds were able to germinate at higher temperatures as the storage temperature increased (compare columns in Fig. 1).

Fig. 1.
Observed germination time-course curves for L. arvense seeds after-ripened at 5, 15 or 24 °C for (A) 0, (B–D) 100 or (E–G) 210 d and subsequently incubated at 5, 8, 10, 13, 15, 17 or 20 °C for 30 d as indicated. No germination ...

Median germination rate (reciprocal of time for 50 % subpopulation germination) also increased with after-ripening time and temperature (Fig. 2). Lowest germination rates were observed at an after-ripening temperature of 5 °C (Fig. 2A) while a significant increase was detected for seeds after-ripened at higher temperatures (Fig. 2B, C). As shown in Figs 1 and and2,2, maximum germination percentages and the highest median germination rate were obtained for seeds after-ripened for 210 d at 24 °C. Final germination percentages for these seeds were 85 % at 5 °C, 100 % between 8–15 °C, 77 % at 17 °C, 20 % at 20 °C and 0 % at 25 °C (Fig. 1G), while fastest germination was observed at incubation temperatures of 10, 13 and 15 °C (Fig. 2C).

Fig. 2.
Lithospermum arvense median germination rate (reciprocal of time for 50 % subpopulation germination) regressed as a function of incubation temperature for seeds after-ripened at (A) 5 °C, (B) 15 °C, or (C) 24 °C for 100, 150 or ...

From these results, it is clear that L. arvense after-ripening patterns were characterized by a widening in the thermal-germination range associated with a progressive increase in the maximum germination temperature, as expected for a winter annual species (Foley, 1994; Allen et al., 1995; Baskin and Baskin, 1998).

Goodness-of-fit and statistical analysis of thermal-germination models

The most accurate approach for the simulation of L. arvense germination response was achieved by assuming a normal distribution of both Tb and Tc in the seed population (Model 2; Table 1). An R2 value of 0·956 was obtained after fitting the model to the observed data considering the complete (global) germination thermal range. Highly significant differences were observed between best-fit values (R2 coefficients of Model 2) and the rest of the coefficients obtained under different models assumptions (Table 1). Residuals and normal probability plots indicated no evidence of lack of homocedasticity or normality in the selected model (data not shown).

Table 1.
Comparison between thermal-germination model outputs obtained after data fitting. Residual sum of squares (SS) and total SS were calculated for the sub-optimal and supra-optimal thermal ranges as well as for the complete germination thermal range (global). ...

From fitting L. arvense subpopulation germination rate to a linear temperature function a normal distribution of both Tb and Tc was inferred in the seed population, as suggested by the selected thermal-germination approach (Table 1). An example of this population attribute is shown in Fig. 3. Germination rates for the different subpopulations obtained from the sub-optimal thermal range of seeds after-ripened at 15 °C for 210 d could be described as a series of positive linear functions with different x-intercepts or Tb(g) values. Similarly, a series of negative linear functions with different Tc(g) values gave an acceptable description of the behaviour of germination rates in the supra-optimal thermal range. The slopes of the regression functions of different population percentiles were relatively similar within each thermal range; thus indicating a small variation in the germination thermal-time estimates among subpopulations (Fig. 3).

Fig. 3.
Germination rates for L. arvense subpopulations regressed as linear functions of incubation temperature. Germination percentiles of 10 % (R2 = 0·74), 20 % (R2 = 0·88), 30 % (R2 = 0·91), 40 % (R2 = 0·91), 50 % (R2 = 0·93), ...

Lithospermum arvense germination rates for the different fractions of the seed population were also fitted to quadratic and Gaussian functions, as a plateau of optimum temperatures was observed at 10 and 20 % percentiles of the population (Fig. 3). In spite of this fact, the most accurate description of the observed germination rates was achieved by the linear regression model (data not shown).

Changes in seed population thermal parameters in relation to after-ripening time and storage temperature

A noticeable increment in the mean maximum germination temperature (Tc(50)) was observed as after-ripening time and storage temperature increased (Table 2; Fig. 4A). The estimated temperature increment between recently harvested seeds and those after-ripened at 24 °C for 210 d was 8 °C. Conversely, the mean base temperature for germination (Tb(50)) showed no consistent variation when regressed as a linear function of the after-ripening time (P = 0·68, R2 = 0·02) and storage temperature (P = 0·42, R2 = 0·07). In the sub-optimal temperature range, lower germination thermal-time (θ1) values were predicted as storage temperature increased (P < 0·01, R2 = 0·57), irrespective of the after-ripening time. Average θ1 values were reduced by 17·6, 34·6 and 40·4 °Cd when seeds were after-ripened at 5, 15 and 24 °C, respectively, compared to recently harvested seeds (Table 2). Observed variations in σTb and σTc values, as well as changes in thermal-time estimates for the supra-optimal thermal range (θ2) were not related to the after-ripening time (P ≥ 0·34).

Fig. 4.
Estimated values of the mean maximum-germination temperature (Tc(50)) for recently harvested seeds of L. arvense and for seeds after-ripened at 5, 15 or 24 °C as indicated, plotted against (A) days of storage, and (B) against after-ripening thermal-time ...
Table 2.
Estimated population thermal parameters for L. arvense seeds after-ripened at 5, 15 or 24 °C. Seeds were incubated at 5, 8, 10, 13, 15, 17, 20 or 25 °C after 0, 45, 100, 150 or 210 d of dry storage. Parameters were obtained by simulation ...

Development of dormancy model

The process of seed dormancy release in L. arvense was characterized by progressive changes in seed population parameters (Tc(50) and θ1) as a function of after-ripening time and storage temperature. In addition, estimated Tc(50) values (Fig. 4A) were higher as after-ripening time and storage temperature increased.

In order to quantify the effect of temperature on seed population dormancy status, changes in Tc(50) were predicted as a function of the accumulation of after-ripening thermal-time units (°Cd) above a ‘base’ temperature for dormancy loss to occur, according to the following equation:

equation image
9

where [minus sign in circle]AT is the thermal-time requirement for after-ripening, Ts is the storage or after-ripening temperature, Tl is the lowest or base temperature (at or below which after-ripening does not occur), and tar is the time required for after-ripening.

As observed in Fig. 4B, changes in Tc(50) values were accurately described by a quadratic function of the after-ripening thermal-time. Optimal Tl was obtained by an iterative procedure until the best fit of the polynomial regression between Tc(50) and [minus sign in circle]AT was attained. Maximum data fit was achieved by considering a Tl value of −6 °C according to the following equation:

equation image
10

Changes in θ1 values were also observed during the process of seed dormancy release (Fig. 5A). Lower values of θ1 were recorded for seeds after-ripened at 15 and 24 °C compared to seeds stored at 5 °C after 100, 150 and 210 d of storage. To account for the effect of [minus sign in circle]AT on θ1, a nonlinear regression analysis was performed. The presence of an outlier value was detected, and thus it was excluded from the definitive data analysis (data not shown). The results presented in Fig. 5B indicate that the reduction in θ1 values was a positive function of the accumulation of after-ripening thermal-time and storage temperature increase. Both variables were combined into a single expression of the form: ln[([minus sign in circle]AT/100) + 1]Ts. An accurate description of the changes in θ1 values was achieved by means of the following quadratic function:

equation image
11

Fig. 5.
Estimated values of the sub-optimal germination thermal-time (θ1) for recently harvested seeds of L. arvense and for seeds after-ripened at 5, 15 or 24 °C as indicated, plotted against (A) days of storage, and (B) against ln[([minus sign in circle] ...

Model performance

Model performance was evaluated using emergence data from an independent experiment. The model was tested under field conditions in order to assess its relative accuracy in the prediction of L. arvense emergence patterns in the region under study. Lithospermum arvense seed lots were exposed to different after-ripening temperatures in the laboratory and the field (see above), and the model was used to estimate the changes in Tc(50) and θ1 values using eqns (10) and (11) according to [minus sign in circle]AT accumulation (eqn 9) and after-ripening temperature. The total accumulation of after-ripening thermal-time units (°Cd) for each seed lot was the sum of the thermal units accumulated by after-ripening in the laboratory and the field. Accumulation of [minus sign in circle]AT began when the daily mean soil temperature recorded in the experimental field at a seeding depth of 1 cm was above the previously determined Tl value (−6 °C). Predicted values of Tc(50) and θ1 for the different seed lots were used to simulate germination dynamics in the field by assuming a normal distribution of both Tb and Tc in the population. Equations (4) and (5) were combined and a seed viability term was added:

equation image
12

where Gu is the proportion of non-viable seeds at the time of seeding (Scott et al., 1984); Tb(50), σTb, σTc and θ2 were considered constant for the simulation process. The following average values were used: Tb(50) = 1·8 °C, σTb = 3·8 °C, σTc = 3·0 °C, and θ2 = 44·0 °Cd. A Gu value of 0·3 was used in the model and the estimation of this coefficient was obtained by performing a tetrazolium chloride viability test (see germination test, above) on samples of 50 seeds obtained from each seed lot prior to seeding (seed viability = 70 ± 6 %, n = 3).

The simulated field germination dynamics and the observed emergence patterns are shown in Fig. 6. The simulation process was started when field temperatures were proximate to entering the permissive range for germination of L. arvense seeds. Dormancy model outputs for seed lot 1 ([minus sign in circle]AT = 1787·3 °Cd; Fig. 6A), seed lot 2 ([minus sign in circle]AT = 2426·7 °Cd; Fig. 6B) and seed lot 3 ([minus sign in circle]AT = 3220 °Cd; Fig. 6C) were Tc(50) = 15·54, 16·65 and 17·69 °C, respectively; estimated values for θ1 were 76·44, 72·3 and 70·53 °Cd, respectively.

Fig. 6.
Predicted field germination percentages and observed emergence data for L. arvense seed lots exposed to different after-ripening treatments in the laboratory and in the field. (A) Seed lot 1 was after-ripened at 20 °C for 33 d and buried in the ...

A fairly acceptable description of the observed emergence patterns was achieved by the simulation model for the different seed lots (Fig. 6). However, the model clearly over-estimated germination percentages at the beginning of the field germination ‘time window’ (April 2005) for seed lot 1 (Fig. 6A) and seed lot 2 (Fig. 6B), and also for the predicted maximum germination plateau in the former. Higher simulated germination values were obtained at the beginning of the evaluated time period, as seeds accumulated after-ripening thermal-time and were progressively released from dormancy. A similar trend was observed for emergence data, as evidenced by a faster rate of seedling emergence in seed lot 3 (Fig. 6C).

Field emergence data was also simulated by assuming a constant Tb value and a normal distribution of θ1 in order to determine whether the predictability of the model could be improved by Model 1 assumptions. A significantly better fit to the field data (F-value = 4·93, P < 0·05) was obtained by assuming a normal distribution of Tb (R2 = 0·73) compared with a constant value of Tb (R2 = 0·66).

DISCUSSION

Based on the results observed, we found that the most accurate thermal-time approach for the simulation of germination response of L. arvense to constant temperatures was achieved by assuming a normal distribution of both base and maximum germination temperatures (Tb and Tc) in the seed population (R2 = 0·956; Table 1). According to this approach, each fraction of the seed population will germinate within a specific temperature range delimited by a unique set of cardinal temperatures. Thus, the probability of germination for each subpopulation will result from the combination of the cumulative normal distributions of both Tb and Tc (eqn 12).

Our results contradict the widely accepted assumption of a single Tb value for the entire seed population. Although many authors have assumed Tb as a unique value, it should be noted that most of these germination models were developed for crop cultivars (Garcia-Huidobro et al., 1982; Covell et al., 1986; Gummerson, 1986; Dahal et al., 1990) in which Tb has been reported as a constant attribute (Bradford, 1995). Conversely, in non-crop species previous reports have also indicated intra-population variation and lack of stability in Tb values (Pritchard et al., 1999; Forcella et al., 2000; Wang et al., 2004; Hardegree, 2006). This situation could be associated with a distinct genetic variability within the seed population, as proposed for Setaria viridis (Wang et al., 1995) and Eurotia lanata (Wang et al., 2004). An alternative explanation could be related to the effect that a heterogeneous maternal environment during seed maturation might have on the seed population response to temperature (Baskin and Baskin, 1998).

From an ecological perspective, the different physiological responses to temperature observed among subpopulations of L. arvense might be considered as an adaptive strategy, where the different fractions of the population accumulate thermal-time at distinct rates, spreading germination over time and thus increasing the probability of seedling establishment.

Some of the estimations of the standard deviations of the normal distribution of Tb predicted extremely low values of this population thermal parameter (Table 2). This might be related to the way in which the mathematical model applied calculates Tb values for each percentile of the population. In fact, even very small changes in the germination rates of the different subpopulations at low incubation temperatures are likely to have a significant influence on estimates of Tb, as observed in the present case.

In the present work, an after-ripening thermal-time model for dormancy release of L. arvense seeds was developed. Model development consisted of the quantification of the observed changes in the optimal thermal-germination parameters previously obtained by assuming a normal distribution of both cardinal temperatures (Tb and Tc) within the population, and the derivation of equations relating the rate of change of each parameter (Tc(50) and θ1) to the after-ripening thermal-time accumulation and storage temperature. Changes in Tc(50) values were adequately described by a quadratic function of [minus sign in circle]AT (Fig. 4B; eqn 10), indicating that the increment in the mean maximum-germination temperature was positively related to the accumulation of [minus sign in circle]AT. Thus, it could be inferred that the Tc(50) rate of increase is a positive function of storage temperature that regulates the rate of after-ripening thermal-time accumulation. These results concur with previous reports that found that the rate of dormancy release was positively related to the after-ripening temperature in winter annual species (Baskin and Baskin, 1976; Foley, 1994; Allen et al., 1995; Bauer et al., 1998; Steadman et al., 2003a, b).

Observed changes in θ1 with after-ripening time (Fig. 5A) were described as a function of [minus sign in circle]AT and Ts (Fig 5B; eqn 11). The relationship found between θ1 and ln[([minus sign in circle]AT/100) + 1]Ts clearly indicates that changes in θ1 depend not only on the accumulation of [minus sign in circle]AT, but also on the temperature experienced by the seeds during the after-ripening process. Thus, it might by inferred that θ1 changes during L. arvense seed dormancy release depend on the amount of accumulated thermal-time units (°Cd) and the rate of [minus sign in circle]AT accumulation, which depends on Ts, as previously mentioned. A similar equation was developed by Batlla and Benech-Arnold (2003) for Polygonum aviculare seeds to describe changes in the standard deviation of the lower-limit germination temperature as a function of stratification temperature and thermal-time accumulation.

An after-ripening base temperature value of −6 °C gave an accurate description of L. arvense seed dormancy loss (Figs 4B and 5B). Available evidence is scarce relating to the effects of sub-zero temperatures during seed dormancy release. Sivakumar et al. (2006) observed a significant reduction in the levels of primary physiological dormancy of Strychnos nux-vomica seeds after exposure to sub-zero temperatures during dry storage. Experimental evidence of the existence of seed metabolic activity at sub-zero incubation temperatures was provided by Wang et al. (2004), who found that E. lanata seeds were capable of accumulating thermal-time and thus germinating when imbibed at temperatures close to −1 °C. The potential occurrence of after-ripening at such low temperatures should make researchers aware of the thermal conditions in which seeds should be stored in order not to alter their dormancy level.

The molecular mechanisms of after-ripening are still not known (Finch-Savage and Leubner-Metzger, 2006); however, the physical properties of the membrane lipids are apparently involved in the perception of temperature by seeds (Hallett and Bewley, 2002). Steadman et al. (2003b) suggested that the higher dormancy-release rates observed at increasing after-ripening temperatures in Lolium rigidum seeds are related to increased membrane fluidity and inherent enzyme temperature–activity relationships promoted by warmer temperatures. An analogous mechanism may act to mediate the role of temperature in the regulation of L. arvense after-ripening process.

The after-ripening thermal-time model developed in this study was capable of predicting the ‘time window’ for L. arvense field germination and the fraction of the seed population that will germinate within this time period, as evidenced by previous seed-bank studies performed by Chantre et al. (2009). Simulated field-germination curves and observed emergence patterns clearly indicate that as L. arvense seeds are progressively released from dormancy, as [minus sign in circle]AT is accumulated, the proportion of germinating seeds and emerged seedlings will increase at the beginning of the field germination ‘time window’ in early autumn (Fig. 6). This is mainly due to a widening in the germination-permissive thermal range associated with an increment in the mean maximum-germination temperature, as expected for a winter annual species (Baskin and Baskin, 1998), and a reduction in the thermal-time requirement for germination at sub-optimal temperatures.

The model performance for predicting L. arvense emergence patterns was quite acceptable considering that seedling emergence involves additional post-germination seedling growth compared with seed germination in the laboratory. In addition, possible fluctuations in the seed water content due to sporadic rainfall events during the after-ripening period in the field could have altered the rate of dormancy release, as suggested by previous reports in other species (Leopold et al., 1988; Foley, 1994; Steadman et al., 2003b; Bair et al., 2006).

From a modelling point of view, we have found strong evidence to support the development of a germination model and an after-ripening thermal-time model for L. arvense based on the concept of a normal distribution of Tb in the seed population. Such evidence was provided by a better fit of the germination response of L. arvense to the applied model (Table 1) and a more accurate prediction of the field emergence data in comparison to the assumption of a constant Tb. However, it should be noted that some of the variation observed in Tb might be associated with the method of data analysis, and also with the unavoidable difficulty in correctly estimating germination rates at low incubation temperatures. Thus, further investigations should be performed in order to increase the predictive capability of the developed model to account for variations in the above mentioned factors.

ACKNOWLEDGMENTS

This research was financially supported by grants from the Universidad Nacional del Sur (PGI 24/A122), the Agencia Nacional de Promoción Científica y Técnica de la Argentina (PICTO-UNS 20032) and the Comisión de Investigaciones Científicas de la Provincia de Buenos Aires (res. 1114/05). G. Chantre held a CONICET (Argentina) postgraduate scholarship.

LITERATURE CITED

  • Allen PS, Meyer SE, Beckstead J. Patterns of seed after-ripening in Bromus tectorum L. Journal of Experimental Botany. 1995;46:1737–1744.
  • Alvarado V, Bradford KJ. A hydrothermal time model explains the cardinal temperatures for seed germination. Plant, Cell and Environment. 2002;25:1061–1069.
  • Bair NB, Meyer SE, Allen PS. A hydrothermal after-ripening time model for seed dormancy loss in Bromus tectorum L. Seed Science Research. 2006;16:17–28.
  • Baskin CC, Baskin JM. Seeds: ecology, biogeography, and evolution of dormancy and germination. San Diego, CA: Academic Press; 1998.
  • Baskin JM, Baskin CC. High temperature requirement for afterripening in seeds of winter annuals. New Phytologist. 1976;77:619–624.
  • Baskin JM, Baskin CC. Temperature requirements for after-ripening in seeds of nine winter annuals. Weed Research. 1986;26:375–380.
  • Batlla D, Benech-Arnold RL. A quantitive analysis of dormancy loss dynamics in Polygonum aviculare L. seeds: development of a thermal time model based on changes in seed population thermal parameters. Seed Science Research. 2003;13:55–68.
  • Bauer MC, Meyer SE, Allen PS. A simulation model to predict seed dormancy loss in the field for Bromus tectorum L. Journal of Experimental Botany. 1998;49:1235–1244.
  • Bell DT. Germination responses to variations in light quality of eight species from sandy habitats in Western Australia. Australian Journal of Botany. 1993;41:321–326.
  • Benech-Arnold RL, Ghersa CM, Sánchez RA, Insausti P. Temperature effects on dormancy release and germination rate in Sorghum halepense (L.) Pers. seeds: a quantitive analysis. Weed Research. 1990;30:81–89.
  • Benech-Arnold RL, Sánchez RA, Forcella F, Kruk BC, Ghersa CM. Environmental control of dormancy in weed seed banks in soil. Field Crops Research. 2000;67:105–122.
  • Bradford KJ. Water relations in seed germination. In: Kigel J, Galili G, editors. Seed development and germination. New York: Marcel Dekker; 1995. pp. 351–396.
  • Bradford KJ. Population-based models describing seed dormancy behaviour: implications for experimental design and interpretation. In: Lang GA, editor. Plant dormancy: physiology, biochemistry, and molecular biology. Wallingford, UK: CAB International; 1996. pp. 313–339.
  • Chantre GR, Sabbatini MR, Orioli GA. Effect of burial depth and soil water regime on the fate of Lithospermum arvense seeds in relation to burial time. Weed Research. 2009;49:81–89.
  • Chatterton NJ, Kadish AR. A temperature gradient germinator. Agronomy Journal. 1969;61:643–644.
  • Covell S, Ellis RH, Roberts EH, Summerfield RJ. The influence of temperature on seed germination rate in grain legumes. I. A comparison of chickpea, lentil, soybean, and cowpea at constant temperatures. Journal of Experimental Botany. 1986;37:705–715.
  • Dahal P, Bradford KJ, Jones RA. Effects of priming and endosperm integrity on seed germination rates of tomato genotypes. I. Germination at suboptimal temperature. Journal of Experimental Botany. 1990;41:1431–1439.
  • Ellis RH, Butcher PD. The effects of priming and natural differences in quality amongst onion seed lots on the response of the rate of germination to temperature and the identification of the characteristics under genotypic control. Journal of Experimental Botany. 1988;39:935–950.
  • Ellis RH, Covell S, Roberts EH, Summerfield RJ. The influence of temperature on seed germination rate in grain legumes. II. Intraspecific variation in chickpea at constant temperatures. Journal of Experimental Botany. 1986;37:1503–1515.
  • Ellis RH, Simon G, Covell S. The influence of temperature on seed germination rate in grain legumes. III. A comparison of five faba bean genotypes at constant temperatures using a new screening method. Journal of Experimental Botany. 1987;38:1033–1043.
  • Favier JF. A model for germination rate during dormancy loss in Hordeum vulgare. Annals of Botany. 1995;76:631–638.
  • Finch-Savage WE, Leubner-Metzger G. Seed dormancy and the control of germination. New Phytologist. 2006;171:501–523. [PubMed]
  • Foley ME. Temperature and water status of seed affect afterripening in wild oat (Avena fatua) Weed Science. 1994;42:200–204.
  • Forcella F, Benech-Arnold RL, Sánchez R, Ghersa CM. Modelling seedling emergence. Field Crops Research. 2000;67:123–139.
  • Garcia-Huidobro J, Monteith JL, Squire GR. Time, temperature and germination of pearl millet (Pennisetum typhoides, S & H). I. Constant temperature. Journal of Experimental Botany. 1982;33:288–296.
  • Grundy AC, Phelps K, Reader RJ, Burston S. Modeling the germination of Stellaria media using the concept of hydrothermal time. New Phytologist. 2000;148:433–444.
  • Gummerson J. The effects of constant temperatures and osmotic potentials on the germination of sugar beet. Journal of Experimental Botany. 1986;37:729–741.
  • Hallett BP, Bewley JD. Membranes and seed dormancy: beyond the anaesthetic hypothesis. Seed Science Research. 2002;12:69–82.
  • Hardegree SP. Predicting germination response to temperature. I. Cardinal-temperature models and sub-population specific regression. Annals of Botany. 2006;97:1115–1125. [PMC free article] [PubMed]
  • Hardegree SP, Van Vactor SS, Pierson FB, Palmquist DE. Predicting variable-temperature response of non-dormant seeds from constant-temperature germination data. Journal of Range Management. 1999;52:83–91.
  • International Seed Testing Association. International rules for seed testing. Seed Science and Technology. 1999;27(Supplement)
  • Karssen CM. Seasonal patterns of dormancy in weed seeds. In: Khan AA, editor. The physiology and biochemistry of seed development, dormancy and germination. Amsterdam: Elsevier Biomedical Press; 1982. pp. 243–270.
  • Leopold AC, Glenister R, Cohn MA. Relationship between water content and afterripening in red rice. Physiologia Plantarum. 1988;74:659–662.
  • Murdoch AJ, Ellis RH. Longevity, viability and dormancy. In: Fenner M, editor. Seeds – the ecology of regeneration in plant communities. Wallingford, UK: CAB International; 2000. pp. 183–214.
  • Murdoch AJ, Roberts EH, Geodert CO. A model for germination responses to alterning temperatures. Annals of Botany. 1989;63:97–111.
  • Pritchard HW, Steadman KJ, Nash JV, Jones C. Kinetics of dormancy release and the high temperature germination response in Aesculus hippocastanum seeds. Journal of Experimental Botany. 1999;50:1507–1514.
  • Schütz W, Milberg P, Lamont BB. Seed dormancy, after-ripening and light requirements of four annual Asteraceae in south-western Australia. Annals of Botany. 2002;90:707–714. [PubMed]
  • Scott SJ, Jones RA, Williams WA. Review of data analysis methods for seed germination. Crop Science. 1984;24:1192–1199.
  • Sivakumar V, Anandalakshmi R, Warrier RR, et al. Effects of presowing treatments, desiccation and storage conditions on germination of Strychnos nux-vomica seeds, a valuable medicinal plant. New Forests. 2006;32:121–131.
  • Sokal RR, Rohlf JF. Biometry: the principles and practice of statistics in biological research. 2nd edn. New York: W.H. Freeman and Company; 1981.
  • Steadman KJ, Bignell GP, Ellery AJ. Field assessment of thermal after-ripening time for dormancy release prediction in Lolium rigidum seeds. Weed Research. 2003;a 43:458–465.
  • Steadman KJ, Crawford AD, Gallagher RS. Dormancy release in Lolium rigidum seeds is a function of thermal after-ripening time and seed water content. Functional Plant Biology. 2003;b 30:345–352.
  • Steinmaus SJ, Prather TS, Holt JS. Estimation of base temperatures for nine weed species. Journal of Experimental Botany. 2000;51:275–286. [PubMed]
  • Thompson PA. Seed germination in relation to ecological and geographical distribution. In: Heywood VA, editor. Taxonomy and ecology. London: Academic Press; 1973. pp. 93–119.
  • Vohnout KD. Mathematical modeling for systems analysis in agricultural research. 1st edn. Amsterdam: Elsevier Science; 2003.
  • Wang R, Bai Y, Tanino K. Effect of seed size and sub-zero imbibition-temperature on the thermal time model of winterfat (Eurotia lanata (Pursh) Moq.) Environmental and Experimental Botany. 2004;51:183–197.
  • Wang RL, Wendel JL, Dekker JH. Weedy adaptation in Setaria spp. I. Isozyme analysis of genetic diversity and population genetic structure in Setaria viridis. American Journal of Botany. 1995;82:308–317.

Articles from Annals of Botany are provided here courtesy of Oxford University Press