Weed seed is photoblastic

Co-adaptation of seed dormancy and flowering time in the arable weed Capsella bursa-pastoris (shepherd’s purse)

The duration of the plant life cycle is an important attribute that determines fitness and coexistence of weeds in arable fields. It depends on the timing of two key life-history traits: time from seed dispersal to germination and time from germination to flowering. These traits are components of the time to reproduction. Dormancy results in reduced and delayed germination, thus increasing time to reproduction. Genotypes in the arable seedbank predominantly have short time to flowering. Synergy between reduced seed dormancy and reduced flowering time would create stronger contrasts between genotypes, offering greater adaptation in-field. Therefore, we studied differences in seed dormancy between in-field flowering time genotypes of shepherd’s purse.


Genotypes with early, intermediate or late flowering time were grown in a glasshouse to provide seed stock for germination tests. Secondary dormancy was assessed by comparing germination before and after dark-incubation. Dormancy was characterized separately for seed myxospermy heteromorphs, observed in each genotype. Seed carbon and nitrogen content and seed mass were determined as indicators of seed filling and resource partitioning associated with dormancy.

Key Results

Although no differences were observed in primary dormancy, secondary dormancy was weaker among the seeds of early-flowering genotypes. On average, myxospermous seeds showed stronger secondary dormancy than non-myxospermous seeds in all genotypes. Seed filling was similar between the genotypes, but nitrogen partitioning was higher in early-flowering genotypes and in non-myxospermous seeds.


In shepherd’s purse, early flowering and reduced seed dormancy coincide and appear to be linked. The seed heteromorphism contributes to variation in dormancy. Three functional groups of seed dormancy were identified, varying in dormancy depth and nitrate response. One of these groups (FG-III) was distinct for early-flowering genotypes. The weaker secondary dormancy of early-flowering genotypes confers a selective advantage in arable fields.

Keywords: Co-adaptation, Capsella bursa-pastoris, shepherd’s purse, seedbank, flowering time, seed heteromorphism, light, nitrate, primary dormancy, secondary dormancy, weed, myxospermy, mucilage


The production of annual, semelparous crops results in an in-field environment that is dominated by regular disturbance by tillage and pesticide application. In addition to competing with the crop for resources, wild plant populations persisting in this environment have to develop sufficiently rapidly following emergence in order to reach maturity and reproduce between disturbance events. In this context, Capsella bursa-pastoris (shepherd’s purse) is developed as a model species for adaptation in situ (Hawes et al., 2005) and a potential biomonitor for sustainable arable production (Karley et al., 2008; Iannetta et al., 2010; Iannetta, 2010; Begg et al., 2011). A main feature of the population structure of C. bursa-pastoris is the prevalence of within-field genotypes that have a genetically distinct short time to reproduction, which are six-fold more common than later reproducing forms (Iannetta et al., 2007). For such annual and self-fertile species, the timing of reproduction is associated with two primary traits, namely ‘time to germination’ from dispersal to completion of germination and ‘time to flowering’ between germination and the onset of flowering. Theoretical studies of plant community ecology suggest the traits should co-evolve (Ritland, 2004) and have identified that differences in time to reproduction underpin the coexistence of functionally distinct plant types (Pachepsky et al., 2001, 2007; Bown et al., 2007). However, there is no empirical evidence for co-adaptation of time to germination and time to flowering.

Time to germination is strongly affected by seed dormancy, which is a complex trait. Seed dormancy is a mechanism that has evolved to ensure that seed germination and plant establishment are spread in time, while dormancy release is controlled by environmental factors that ensure synchronization with the optimal growing season (Baskin and Baskin, 1989). Different classes of seed dormancy are distinguished (Baskin and Baskin, 1998). Among members of the Brassicaceae, physiological dormancy is common, characterized by the absence of axis elongation (or radicle protrusion) under conditions that are otherwise favourable for germination (Bewley and Black, 1994; Finch-Savage and Leubner-Metzger, 2006). Two types of physiological dormancy are distinguished: primary and secondary. Primary dormancy is observed in seeds upon dispersal from the mother plant. Secondary dormancy occurs in seeds which have lost their primary dormancy, but which in the absence of completion of germination acquire a second dormancy phase (Hilhorst, 1995). A temperature that is not optimal could result in inhibition of germination and ultimately in the induction of secondary dormancy. Another factor that can induce secondary dormancy in positive photoblastic seeds is the duration of dark incubation, which is widely important for seeds in agricultural soils, as it occurs for example when seeds are buried in the soil. Dark incubation of imbibed seeds at a constant temperature, followed by exposure to light and recording of subsequent germination, has been shown to be an effective method in the laboratory to study differences in secondary dormancy between genotypes of Arabidopsis thaliana (Toorop et al., 2005; Cadman et al., 2006). This method should also be applicable to C. bursa-pastoris, given the small phylogenetic distance between these species (Bailey et al., 2006; Finch-Savage and Leubner-Metzger, 2006).

Examination of the different forms of dormancy in C. bursa-pastoris necessitates study of a wider range of factors involved in germination and dormancy release. Both light and alternating temperatures have been regarded to induce germination in primary dormant seeds, acting together for completion of germination of seeds on or close to the soil surface (Popay and Roberts, 1970; Froud-Williams et al., 1984). Nitrate was demonstrated to effectively induce germination in dormant seeds of C. bursa-pastoris, but only at alternating temperatures (Popay and Roberts, 1970). Similar to the response to light, the response to nitrate (Pons, 1989) and to alternating temperatures (Thompson et al., 1977) can play a role in gap detection in the vegetation, facilitating establishment of the seedling by avoiding competition, e.g. in arable fields. Presumably, these same factors also play a role in germination and dormancy of seeds buried in the soil and of secondary dormant seeds in the laboratory. A. thaliana seeds that were made secondary dormant by dark incubation of imbibed seeds at 20 °C developed the requirement for nitrate for completion of germination (Cadman et al., 2006). The gradual change in seed dormancy that is apparent as a shift in sensitivity to nitrate and light during dry after-ripening indicates that these environmental factors are related to dormancy release and induction of germination, affecting the expression of several hundred genes that are associated with either dormancy or germination in a quantitative and similar manner (Finch-Savage et al., 2007).

The variation in the different forms of physiological dormancy among field populations of C. bursa-pastoris is uncertain, but the range of dormancy and germination characteristics in this species is potentially widened and complicated by the presence of myxospermy, in which upon contact with water, seeds produce mucilage. This phenomenon is thought to play a role in adhesion of seeds to the soil, facilitating a suitable environment for offspring near the parent plant (Gutterman and Shem-Tov, 1997), to allow the formation of a hydrogel around the seed in an environment with low osmotic potential (Deng et al., 2012), and to promote DNA repair (Huang et al., 2008). Shepherd’s purse plants produce both myxospermous and non-myxospermous seeds on the same plant. The role of myxospermy in dormancy has been suggested, although it has not been described (for discussion see Toorop et al., 2008).

Various biological and environmental factors interact to influence dormancy and its release and thereby the time to germination in the reproductive context. As both time to germination and time to flowering are major contributors to timing of reproduction, it can be expected that these two traits are co-adapted, acting in conjunction to enhance the contrast in timing of reproduction. Therefore, the objective of this study was to investigate if reduced seed dormancy and early time to flowering in genotypes of shepherd’s purse concurred. A selection of genotypes from arable fields were used that had been previously characterized for time to flowering (Iannetta et al., 2007). Because of the possible role of myxospermy in seed dormancy and germination, myxospermous and non-myxospermous seeds were evaluated separately. As genotypes differ in somatic input and reproductive output, the seed carbon and nitrogen content was studied to assess an association with seed filling and resource partitioning in the genotypes, and in the myxospermous and non-myxospermous seeds.


Plant material

Twelve genotypes of Capsella bursa-pastoris (L.) Medik. (accession codes, SCRI-156, -158, -177, -367, -416, -469, -707, -773, -798, -799, -930 and -937) were used, derived originally from soil samples taken during the Farm Scale Evaluations (Champion et al., 2003), and selected out of 52 available genotypes from existing knowledge of their flowering time (Iannetta et al., 2007). Seed samples from each genotype weighing 2 mg were surface-sterilized in 1 mL of 1 % sodium hypochlorite (NaOCl) with 0·01 % (v/v) Tween20™ for 5 min with shaking. After washing (3×) in 1 mL of sterile distilled water the seeds were germinated on sterile distilled water agar (1 %, w/v), stratified for 3 d at 4 °C and cultured at 21 °C. One-week-old seedlings were transferred to small peat pots for 6 weeks before they were transferred to large peat-containing pots. Six sibling plants per genotype were grown to maturity simultaneously, positioned in a random plot design in a glasshouse. These plants, grown in autumn 2006, were considered replicate plants for the genotypes. The plants were provided with water daily (150 mL), and 150 mL of liquid fertilizer [10 % (w/v) ‘20 : 20 : 20 Sangral Soluble Fertiliser’ (William Sinclair Horticulture Ltd, Lincoln, UK)] every seventh day. Plants were provided with 18 h of daylight (>250 W m −2 ) and a day/night temperature regime of 25/15 °C. Plants were bagged before flowering to ascertain self-pollination and for seed collection, which occurred around the same time. Plants were allowed to fully mature and senesce before seeds were harvested. Seeds were harvested separately per replicate plant and we considered seed lots from replicate plants. Seeds were stored at 15 % relative humidity and 15 °C. Categorization of the plants of these genotypes by flowering time (time between sowing and flowering) according to Iannetta et al. (2007) into the following three groups was confirmed: early, intermediate or late time to flowering (TTF), the mean TTF (±s.e.) for each group being 59 ± 3 (n = 3), 80 ± 4 (n = 4) and 113 ± 10 d (n = 5), respectively. On seed lots from each plant, light brown myxospermous and dark brown non-myxospermous seeds were partitioned by eye in a two-step process. First, light- and dark-brown seeds were separated. Each batch was then separated again to isolate the lightest- and darkest-brown seeds. Of the four resultant seed fractions the two extreme seed colour fractions were used to acquire data. Thus, the experiments comprised 12 genotypes, three (for germination and dormancy studies) or six (for carbon and nitrogen content) plant replicates per genotype, and two seed colour fractions per plant replicate.


Germination was performed in separate 5-cm Petri dishes with two filter papers (Whatman no. 1) and wetted with either 1·5 mL demineralized water or 10 m m KNO3 (Fisher Scientific, Loughborough, UK) or 100 µ m fluridone (Sigma-Aldrich, St Louis, MO, USA; Grappin et al., 2000). Primary dormancy was assessed at the optimal temperature by comparing germination in water with germination in nitrate and fluridone. For each replicate plant and treatment, 90 seeds were used, divided equally between three replicate dishes. To determine the optimum germination temperature, seeds were incubated at either 25 °C or at day/night temperatures of 30/20, 25/15, 20/10, 15/5, 30/15, 25/10 or 20/5 °C, performed with one plant replicate of genotype SCRI-177. The optimum temperature was used for germination in subsequent experiments. A white light regime of 8 h during the day was applied, unless stated otherwise.

To test the capacity for secondary dormancy and the extent to which secondary dormancy differed between the flowering time categories, seeds of all genotypes were imbibed on water and incubated in darkness at a constant temperature of 25 °C to induce secondary dormancy prior to transfer to 25/10 °C with white light for germination. Dark-incubation was applied by wrapping dishes in two layers of aluminium foil and placing them inside a black plastic bag inside the incubator. Dormancy depth was defined as the difference in germination with and without prior incubation of imbibed seeds for 18 d in darkness (i.e. reduction in germination), interpreted as a reflection of dormancy after excluding loss of viability as a possible cause for this reduction. It was assumed that seeds with the capacity for stronger induced dormancy will reach this dormancy more rapidly than seeds with the capacity for weaker induced dormancy, which is reflected in lower germination after a fixed dark-incubation time. Confirmation of dormancy was provided by induction of germination through application of 10 m m KNO3. The response to nitrate of secondary dormant seeds was calculated for each within-genotype replicate by subtracting the individual value for percentage germination in water from the value in nitrate.

Completion of germination in the light, defined as protrusion of the radicle >1 mm through the enveloping seed layers, was scored at regular intervals. Germination in darkness was assessed at the end of the dark-incubation. Germination speed or time to half-maximal germination (t50) was determined after sigmoidal curve fitting using the Boltzmann equation: y = A2 + (A1 – A2)/(1 + exp((xx0)/dx)) (Origin v6·1). After optimization, controlled deterioration tests were performed by exposing the seeds to a relative humidity of 70 % at 45 °C for 14 d, by storing them above a 24 % LiCl solution in a sealed light-free box (Hay et al., 2008). These tests were intended to produce intermediate viability, allowing comparison with greater discriminatory power between genotypes due to the sigmoidal nature of the decay curve. Viability assessment through tetrazolium staining was performed according to the International Seed Testing Association (1996).

Carbon and nitrogen content

Carbon and nitrogen content were determined to assess seed filling and resource partitioning in association with seed dormancy. Six replicate samples comprising five (whole and weighed) myxospermous or non-myxospermous seeds were weighed and processed for the four different genotypes within each of the three TTF categories (early, intermediate and late). The 144 samples (48 per TTF class, 72 per seed category) were contained within tin containers for combustion and processing with a CE440™ Elemental Analyser according to the manufacturer’s recommendations (Exeter Analytical, Coventry, UK).

Statistical analysis

Before analysis, the germination fraction variables were normalized by arcsine transformation. A mixed-effect ANOVA was used to test the fixed effect of germination conditions and presence of myxospermy, and TTF on the various seed quality and germination variables. Grouping of observations by genotype (nested within TTF) was treated as a random effect. In addition, the effect of myxospermy and TTF on the germination response to nitrate treatment of seeds at varying depths of dormancy was tested by mixed-effects ANCOVA. As before, the presence of myxospermy and TTF were treated as fixed, with the addition of the nitrate-related covariate (KNO3) plus the inclusion of genotype as a random effect on both slope and intercept. Given the relative complexity of the resulting model, the interactive effect of presence of myxospermy and TTF on the response to the nitrate–dormancy depth relationship was explored by fitting a series of models, first testing the effect of presence of myxospermy separately for each TTF category and then the effect of TTF for each of the myxospermy categories. Sigmoidal relationships were captured by logistic regression in which the response variable, here the germination fraction under nitrate treatment, is related to a linear combination of treatment effects by the logit link function. The normalizing transformation of the response variable is not required in this case. Tests were performed with TTF and presence (or absence) of myxospermy as covariates; and dormancy depth as treatment effect. The genotype groupings were discounted in this analysis. The use of the logit link assumes the response is bounded between zero and one, requiring two observations in which the dormancy depth fraction of less than zero was omitted from the analysis. As with the previous analysis, the interactive effect of presence of myxospermy and TTF on the response to the nitrate–dormancy depth relationship was explored by fitting a series of models testing the effect of presence of myxospermy separately for each TTF category and then the effect of TTF for each of the myxospermy categories. Curve fitting was done using the Boltzmann equation: y = A2 + (A1 – A2)/(1 + exp((xx0)/dx)) (Origin v6·1).


Seed heteromorphism

Both myxospermous (light-brown colour) and non-myxospermous (dark-brown) propagules were apparent in the seed batches of all the genotypes studied. The 100-seed weight was significantly lower in non-myxospermous (10·3 mg) than myxospermous seeds (11·3 mg; F1,57 = 39·4, P < 0·0001), and did not differ between the three TTF categories. The carbon/nitrogen (C : N) ratio was significantly lower in non-myxospermous than in myxospermous seeds (Table 1 ; F1,55 = 19·8, P < 0·001). In addition, absolute levels of N, but not C, differed significantly, with seeds of early-flowering time genotypes containing more nitrogen (F2,52 = 5·9, P = 0·004) per seed than the other later TTF genotypes; and non-myxospermous seeds containing more nitrogen than myxospermous seeds (F1,55 = 16·8, P < 0·001).

Table 1.

Carbon (C) content, nitrogen (N) content and C : N ratio in seeds from Capsella bursa-pastoris of genotypes with early, intermediate or late flowering times

Flowering time Seed character C : N ratio C (% dry mass) N (% dry mass)
Early Myxospermous 11·4±0·1 54·0±0·1 4·75±0·04
Non-myxospermous 10·9±0·1 53·6±0·4 4·91±0·04
Intermediate Myxospermous 11·9±0·4 54·5±0·1 4·59±0·14
Non-myxospermous 10·9±0·3 53·6±0·3 4·91±0·13
Late Myxospermous 12·0±0·3 54·0±0·3 4·50±0·09
Non-myxospermous 11·5±0·1 54·2±0·2 4·71±0·04

Data were generated separately for fractions with and without myxospermy. Values are means ± s.e.

Seed viability was tested through germination testing and tetrazolium staining. Although initial viability differed significantly with myxospermous seeds showing higher germination than non-myxospermous seeds (99·3 and 92·3 %, respectively; F1,57 = 5·5, P = 0·0226), this difference was small. A controlled deterioration test reduced germination to 66·8 %, but germination did not differ significantly between myxospermy or TTF classes (data not shown).

Primary dormancy

Final germination percentage (mean ± s.e.; n = 3 replicates) for Capsella bursa-pastoris (shepherd’s purse) seeds (genotype SCRI-177) at different regimes that compare constant and alternating (day/night) temperatures. Myxospermous and non-myxospermous seeds are as indicated.

Final germination percentage of seeds from different flowering time variants (early, intermediate and late) of Capsella bursa-pastoris (shepherd’s purse). Germination data were gathered from myxospermous and non-myxospermous seeds (as indicated) imbibed in water at alternating temperatures (25/10 °C). Data are means ± s.e., n ≥ 3 genotypes, i = 3 plant replicates per genotype.

Secondary dormancy

A dark-incubation period of 18 d was sufficient to induce intermediate levels of secondary dormancy (i.e. 39 %; Fig. 3 ), observed as a reduced germination response, and was used in subsequent tests for comparison. There was no significant difference in germination during the dark-incubation between the heteromorphic seed types or TTF classes (F1,57 = 3·9, P = 0·053; Fig. 4 A). Upon subsequent exposure to optimal germination conditions (light, 25/10 °C), the remaining seeds of the flowering genotypes showed significant differences in germination, with 55·1 % germination for early, 30·4 % for late and 23·8 % for intermediate TTF genotypes (F2,9 = 4·5, P = 0·0444; Fig. 4 B). Germination in the light was higher for non-myxospermous seeds (43·9 %) than for myxospermous seeds (24·8 %; F1,57 = 38·3, P < 0·0001). There were differences between the myxospermy classes for the TTF groups, with 32·7 % light-stimulated germination for early, 15·3 % for intermediate and 14·1 % for late TTF forms (F2,57 = 3·8, P = 0·0295). The induction of secondary dormancy was studied more closely by quantifying the increase in dormancy depth. The TTF genotypes differed in the increase in dormancy depth, with 29·0 % increase for early, 51·7 % for intermediate and 46·0 % for late TTF genotypes (F2,9 = 5·4, P = 0·0289; data not shown). In addition, myxospermous seeds showed significantly stronger secondary dormancy (49·8 %) than non-myxospermous seeds (37·5 %; F1,57 = 14·9, P = 0·0003; Fig. 4 B). In a parallel experiment, dark-incubated seeds transferred to the same conditions (light, 25/10 °C) and 10 m m KNO3 showed that non-myxospermous seeds germinated more (79·9 %) compared with myxospermous seeds (65·9 %; F1,57 = 16·5, P = 0·0002; Fig. 4 C). More seeds completed germination in the presence of nitrate (72·9 %) than in water (34·4 %; t = 15·9, P = 0; compare Fig. 4 B and C) after 18 d darkness.

Final germination percentage for Capsella bursa-pastoris (shepherd’s purse) seeds with or without myxospermy, as indicated. Seeds were germinated at 25/10 °C with white light after varying periods of dark incubation (to induce secondary dormancy) at a constant temperature (25 °C). Data are means ± s.e., n = 12 genotypes, i = 3 plant replicates per genotype.

Germination of seeds from flowering time variants (early, intermediate and late) of Capsella bursa-pastoris (shepherd’s purse): (A) during incubation in the dark at 25 °C (scored after 18 d); and (B, C) after transfer from dark incubation (secondary dormancy induced) to alternating day/night temperature regimes of 25/10 °C and white light in: (B) water or (C) 10 m m KNO3. Myxospermous and non-myxospermous seeds as indicated. Data are means ± s.e., n ≥ 3 genotypes, i = 3 plant replicates per genotype.

Response to nitrate of secondary dormant seeds

Among the TTF genotypes and myxospermy types, non-myxospermous seeds of early TTF genotypes were less responsive to nitrate than seeds of other classes as a result of weaker dormancy (F2,57 = 8·3, P = 0·0007; Fig. 4 B, C). There was an approximate linear relationship between the response to nitrate and the dormancy depth of seeds from all TTF classes (Fig. 5 ). The relationship was also similar (not significant) for each myxospermous seed type among the TTF classes. However, among the intermediate genotypes a significant effect of the presence of myxospermy was observed on the intercept (P = 0·031) and slope (P = 0·020); and although the slope in late-flowering time genotypes was not significant (P = 0·052; data not shown) this provided sufficient scope for further analysis. Among non-myxospermous seeds, the relationship between the response to nitrate and dormancy depth did not differ significantly across the TTF genotypes. However, for myxospermous seeds, there were significant differences between TTF categories in the intercept (P = 0·029) and slope (P < 0·0001) of the relationship: seeds of early-flowering genotypes showing a steeper slope (P < 0·001) and smaller intercept (P < 0·001) than intermediate and late genotypes. There was no significant difference between the intermediate and late TTF genotypes. Myxospermous seeds of the intermediate and late TTF genotypes exhibited a stronger nitrate response and stronger dormancy than the non-myxospermous seeds (Fig. 6 A, B). The non-myxospermous and myxospermous seeds of early TTF genotypes appeared to group together and display even weaker dormancy and lower response to nitrate than the non-myxospermous seeds of intermediate and late TTF genotypes (Fig. 6 C).

Dormancy depth, calculated as the difference in dormancy after subtraction of the percentage seed germination after dark-incubation from that percentage before incubation, versus the response of secondary dormant seeds placed in nitrate (10 m m KNO3, i.e. Fig. 4 C) after 18 d dark incubation. This is shown for seeds that are either (A) myxospermous or (B) non-myxospermous from flowering time variants of Capsella bursa-pastoris (shepherd’s purse). Early-, intermediate-and late-flowering time genotypes are indicated in the keys.

Dormancy depth versus the response of secondary dormant seeds placed in nitrate (10 m m KNO3) after 18 d darkness in individual genotypes and plant replicates of Capsella bursa-pastoris (shepherd’s purse) in myxospermous and non-myxospermous seeds (as indicated) of early-, intermediate- and late-flowering time genotypes (as indicated) with: (A) myxospermous seeds of intermediate- and late-flowering genotypes; (B) non-myxospermous seeds of intermediate- and late-flowering genotypes; (C) myxospermous and non-myxospermous seeds of early-flowering genotypes. Logistic modelequations were y = 154·4 + x – 29·7]/106·4)> (A), y = 81·6 + x – 15·5]/17·7)> (B) and y = 90·4 + x – 34·8]/8·8)> (C).

Functional dormancy groups

Observed differences in seed physiology indicated the existence of distinct dormancy groups based on myxospermy and TTF. When the response to nitrate across the full range of dormancy depth was considered for all seed types, some seed types showed a sigmoidal relationship between the response to nitrate and dormancy depth (Fig. 6 ). This is consistent with the anticipated constraints in the response to nitrate between lower and upper limits of 0 and 100 %, respectively. Logistic regression analysis of the sigmoidal relationship identified differences in the shape of this relationship between the TTF genotypes for myxospermous (P < 0·0001) and non-myxospermous (P = 0·0004) seeds. Repeating these comparisons with the exclusion of the early TTF genotypes revealed no differences between the remaining intermediate and late TTF genotypes for non-myxospermous and myxospermous seeds. The form of the relationship between the response to nitrate and increased dormancy depth differed between the non-myxospermous and myxospermous seeds of the intermediate- and late-flowering genotypes (P < 0·0001), consistent with the apparent homogeneous response to nitrate of myxospermous seeds. In contrast, the relationship of myxospermous and non-myxospermous seeds among the early-flowering genotypes did not differ significantly (P = 0·11). Taken together, three distinct functional groups (FGs) were distinguished, in which the seeds were classified by their relationship between the response to nitrate and dormancy depth into: FG-I, which comprises myxospermous seeds of intermediate and late TTF genotypes (Fig. 6 A); FG-II, non-myxospermous seeds of the same genotypes (Fig. 6 B); and FG-III, both myxospermy classes of early-flowering genotypes (Fig. 6 C).


Our data have demonstrated that primary dormancy was marginal and did not differ between the TTF in-field genotypes of shepherd’s purse (Fig. 2 ). On the other hand, secondary dormancy was obtained easily and was much stronger than primary dormancy (Fig. 3 ). There was weaker secondary dormancy in seeds from early TTF genotypes (Fig. 4 ). Thus, time to germination and time to flowering seem to have co-adapted. The long-lived character of the shepherd’s purse seeds in the soil suggests that genotypic differences in secondary dormancy are more relevant than primary dormancy for the life-history characteristics of this species (Darlington and Steinbauer, 1961). Our principal findings are that the depth of secondary dormancy (and the response to nitrate) localizes the seeds of the early reproducing genotypes to a single, weak dormancy, functional group (FG-III) that is distinct from seeds of the later (intermediate and late) reproducing forms (Fig. 6 ). Two other categories of non-myxospermous (FG-II) and myxospermous (FG-I) seeds are functionally distinct, so that both intermediate- and late-reproducing genotypes possess seeds with weaker (non-myxospermous) and stronger (myxospermous) secondary dormancy. The production of FG-I seeds that express a stronger secondary dormancy could partially mitigate the risks associated with the less structured behaviour of seeds associated with later flowering types. However, the later flowering plant types also produce FG-II, i.e. weaker dormant, seeds. Although these seeds are at risk of not reproducing, as the seedlings produced are potentially all exposed to a single catastrophic event, their presence reduces the negative aspect of dormancy, which is a reduction in mean fitness for the genotype, and provides an opportunity to form a biennial life cycle. Although our results confirm a role for nitrate in induction of germination upon seed dormancy release in shepherd’s purse (Popay and Roberts, 1970), we highlight the importance of intraspecific assessments of functional variation; the depth of dormancy differs between and within TTF genotypes of the species. Secondary dormancy in genotypes of FG-III (the early TTF group) is sufficiently weak to result in germination with light (and water) alone, reflecting a high level of nitrate-independence and making these genotypes physiologically distinct from the later TTF genotypes.

The studied UK genotypes seem to behave differently from those from Sweden, where considerable primary dormancy was observed (Baskin et al., 2004). They also behave differently from a UK accession described by Popay and Roberts (1970), which displayed strong primary dormancy. This difference possibly reflects an adaptation to the different climatic conditions or, alternatively, the reduced primary dormancy in the studied UK genotypes is an adaptation to arable fields. Furthermore, the predominance of the early-flowering genotypes that produce FG-III seeds within arable fields across the UK (Iannetta et al., 2007) suggests that reduced secondary dormancy combined with rapid reproduction is the preferred strategy in this environment. Like time to flowering, dormancy is an important adaptive strategy in weeds faced with environmental stochasticity, allowing populations to spread the risk of future disturbance and avoid annihilation as a consequence of a single, pre-reproduction, field-wide disturbance event (Childs et al., 2010). In genotypes producing the FG-III seeds, the combination of rapid germination prior to the application of nitrogen fertilizer and crop establishment and subsequent early flowering is likely to increase the probability of successful reproduction. Success of this strategy is consistent with a diminished requirement for dormancy. The ‘weediness’ (Baker, 1974) of early-flowering genotypes that combine high fecundity (Iannetta et al., 2007) and weak secondary dormancy with early flowering is consistent with an r-selected life-history strategy that is well adapted to frequent disturbance (Begon et al., 1986). Such r-strategists direct resources to reproduction as opposed to somatic output and this improves the likelihood of persistence in environments that are time limited by major events such as ploughing, herbicide application or harvest. Genetic evidence for the association of seed dormancy, seed fecundity and TTF was provided by Huang et al. (2010) in A. thaliana, a species that is phylogenetically closely related to shepherd’s purse, and was further associated with plant lifetime fitness.

Myxospermy appears to form a component of the dormancy bet-hedging strategy (Rees et al., 2006; Childs et al., 2010). Seed heteromorphism has been described for species in a limited number of families, including the Brassicaceae (Sorensen, 1978). The production of heteromorphic seeds was thought to broaden the range of conditions required for germination and subsequent plant growth, increasing the chances of reproduction (Fenner and Thompson, 2005). Seed heteromorphism was described to coincide with different germination and soil persistence strategies in Atriplex species (Carter and Ungar, 2003). Heteromorphic shepherd’s purse seeds in this study reflect seed classes whose dormancy behaviour is underpinned by particular physiological and biochemical states (Bewley and Black, 1994). A similar colour heteromorphism in Sisymbrium officinale also revealed a non-myxospermous and myxospermous fraction. However, the latter fraction was associated with high quality while the former fraction displayed low viability (Iglesias-Fernandez et al., 2007), which contrasts with the high viability of the non-myxospermopus shepherd’s purse seeds.

Within-field arable weed populations are typically presented with narrow windows for growth that result from competition with crop plants for light and nutrients, and by herbicide and tillage regimes. The difference in 100-seed weight of the myxospermous and non-myxospermous seeds appeared to reflect differences in seed filling, with a lower mass for non-myxospermous seeds. However, the carbon content did not differ between TTF categories or myxospermy types, indicating similar seed filling and maturity. Early-flowering genotypes demonstrated the capacity to partition nitrogen resources to their seeds with greater efficiency than the later flowering forms; and non-myxospermous seeds partitioned more nitrogen than myxospermous seeds (Table 1 ). The trend of increased seed resources being linked to weaker dormancy is also a characteristic of domesticated species and is probably the result of faster seed filling combined with unperturbed development; we may therefore postulate that such attributes are pleiotropic. The counterintuitive higher nitrogen content in seeds with a lower mass is probably due to selectively stronger nitrogen partitioning.

The results reported here provide evidence that flowering time and seed secondary dormancy are synergistic traits in shepherd’s purse. Primary dormancy and flowering time traits have also been linked at the molecular level in the related species A. thaliana (Colucci et al., 2002; Chiang et al., 2009), and Flowering Locus C (FLC) was proposed as a key gene in the control of seed dormancy (Chiang et al., 2009). Flowering time variation in shepherd’s purse has also been attributed to differences in FLC (Slotte et al., 2009). In addition, H2B deubiquitination is involved in the control of seed dormancy (Liu et al., 2007) and flowering through the same gene FLC (Cao et al., 2008; Schmitz et al., 2009). Thus, it seems reasonable to presume that the observed concurrence of variation in flowering time and seed dormancy in shepherd’s purse genotypes is the result of a genetic or epigenetic event. Three functionally distinct clusters were also identified for shepherds purse on the basis of flowering time and other (non-seed) traits which further supports the hypothesis that the genetic control of key life-history traits differs significantly among the intraspecific forms (Iannetta et al., 2007). More work is required to unravel the molecular mechanisms that underlie the link between flowering time, seed dormancy and other life-history traits in shepherd’s purse.

In conclusion, the correlation between seed secondary dormancy and flowering time in the seeds of shepherd’s purse indicates the pivotal role of the earliest life-history stage in the determination of time to reproduction. Weaker dormancy in the seeds of early-flowering forms confers a selective advantage in the extreme selection pressures of the arable environment, and determines plant persistence and coexistence.


This work was supported by Scottish Government funding to the James Hutton Institute (P.P.M.I., G.S.B., G.R.S.); and by grant-in-aid from the UK Department for Environment, Food and Rural Affairs (DEFRA) to the Millennium Seed Bank Project of the Royal Botanic Gardens, Kew (P.T.).

Germination biology of Hibiscus tridactylites in Australia and the implications for weed management

Hibiscus tridactylites is a problematic broadleaf weed in many crops in Australia; however, very limited information is available on seed germination biology of Australian populations. Experiments were conducted to evaluate the effect of environmental factors on germination and emergence of H. tridactylites. Germination was stimulated by seed scarification, suggesting the inhibition of germination in this species is mainly due to the hard seed coat. Germination was not affected by light conditions, suggesting that seeds of this species are not photoblastic. Germination was higher at alternating day/night temperatures of 30/20 °C (74%) and 35/25 °C (69%) than at 25/15 °C (63%). Moderate salinity and water stress did not inhibit germination of H. tridactylites. Seedling emergence of H. tridactylites was highest (57%) for the seeds buried at a 2 cm depth in the soil; 18% of seedlings emerged from seeds buried at 8 cm but no seedlings emerged below this depth. Soil inversion by tillage to bury weed seeds below their maximum depth of emergence could serve an important tool for managing H. tridactylites.


Figure 1. Effect of duration of scarification…

Figure 1. Effect of duration of scarification with sulfuric acid (H 2 SO 4 )…

Figure 1. Effect of duration of scarification with sulfuric acid (H2SO4) on seed germination of Hibiscus tridactylites after 14 days of incubation at 30/20 °C alternating day/night temperature.

The line represents a three-parameter Gaussian model fitted to the data. Maximum germination was achieved when seeds were scarified with H2SO4 for 20 minutes.

Figure 2. Effect of sodium chloride concentration…

Figure 2. Effect of sodium chloride concentration on germination of scarified seeds of Hibiscus tridactylites…

Figure 2. Effect of sodium chloride concentration on germination of scarified seeds of Hibiscus tridactylites after 14 d of incubation in light/dark at 30/20 °C alternating day/night temperature.

The line represents a three-parameter sigmoid model fitted to the data. The sodium chloride concentration required for 50% inhibition of the maximum germination was 193 mM.

Figure 3. Effect of osmotic potential on…

Figure 3. Effect of osmotic potential on germination of scarified seeds of Hibiscus tridactylites after…

Figure 3. Effect of osmotic potential on germination of scarified seeds of Hibiscus tridactylites after 14 d of incubation in light/dark at 30/20 °C alternating day/night temperature.

The line represents a three-parameter sigmoid model fitted to the data. The osmotic potential required for 50% inhibition of the maximum germination was −0.48 MPa.

Figure 4. Effect of burial depth on…

Figure 4. Effect of burial depth on seedling emergence of scarified seeds of Hibiscus tridactylites.

The line represents a three-parameter Gaussian model fitted to the data. Maximum emergence was achieved at the burial depth of 2.0 cm.