Introduction
Astragalus L. (Fabaceae, subfam. Papilionoideae, tribe Galegeae) is the largest genus of seed plants with 3000 or more species (Frodin, Reference Frodin2004; Chaudhary et al., Reference Chaudhary, Rana and Anand2008) that occurs naturally on several continents (Chaudhary et al., Reference Chaudhary, Rana and Anand2008; Mabberley, Reference Mabberley2008). Astragalus has some economic value for use as forage for livestock (Butkutė et al., Reference Butkutė, Padarauskas, Cesevičienė, Taujenis and Norkevičienė2018; Tahmasebi et al., Reference Tahmasebi, Manafian, Ebrahimi, Omidipour and Faal2020), as a medicine (Li et al., Reference Li, Li, Gan, Song, Kuang and Li2013, Reference Li, Qu, Dong, Han, Liu, Fang, Zhang and Wang2014; Liu et al., Reference Liu, Zhao and Luo2017) and in the industry (Lôpez-Franco and Higuera-Ciapara, Reference Lôpez-Franco, Higuera-Ciapara, Phillips and Williams2009; Ferhi et al., Reference Ferhi, Das, Elaloui, Moussaoui and Yanez2014). However, some species of Astragalus are poisonous to livestock (Cook et al., Reference Cook, Ralphs, Welch and Stegelmeier2009, Reference Cook, Gardner, Martinez, Robles and Pfister2017), thus causing economic loss to the livestock grazing industry.
Many Astragalus species are rare and threatened with extinction. For example, NatureServe (2019) lists 81 species of Astragalus in the USA and Canada as imperilled or critically imperilled. The red list report of the International Union for the Conservation of Nature (IUCN) (2019) reported that one species is extinct and 32 endangered or critically endangered. Factors implicated in the rarity and threatened status of Astragalus species include low seed germination percentage and slow seedling growth (Kunz et al., Reference Kunz, Randall, Gray, Wall and Hohmann2016; Statwick, Reference Statwick2016), pollinator (pollen) limitation (Becker et al., Reference Becker, Voss and Durka2011; Baer and Maron, Reference Baer and Maron2018; Schurr et al., Reference Schurr, Affre, Flacher, Tatoni, Pecheux and Geslin2019), urban and land development (Decker, Reference Decker2005, Reference Decker2006), herbivory (Lesica Reference Lesica1995; Dianati Tilaki et al., Reference Dianati Tilaki, Naghipour Borj, Tavakoli and Haidarian Aghakhani2010; Baer and Maron, Reference Baer and Maron2018) and small population size (Decker, Reference Decker2006; Wells, Reference Wells2006).
For Astragalus, sowing seeds (usually scarified) or planting ex situ-produced seedlings/juveniles (usually from scarified seeds but also from stem cutting in the case of A. tennesseensis; Bowles et al., Reference Bowles, Bachtell, DeMauro, Sykora and Bautista1988) in the field can be used for reintroduction of extirpated populations (Bowles, Reference Bowles1988; Bowles et al., Reference Bowles, Betz and DeMauro1993; Erisen et al., Reference Erisen, Yorgancilar, Atalay, Babaoglu and Duran2010; Naseri and Adibi, Reference Naseri and Adibi2016), introduction for the establishment of new populations (Baskin and Baskin, Reference Baskin and Baskin1981; Bowles et al., Reference Bowles, Betz and DeMauro1993; Kondo and Takeuchi, Reference Kondo and Takeuchi2004; Albrecht and McCue, Reference Albrecht and McCue2010; Albrecht and Penagos, Reference Albrecht and Penagos2012; Albrecht and Long, Reference Albrecht and Long2019) or for augmentation of existing populations (Becker, Reference Becker2010). Also, plants can be produced by tissue and cell culture methods. Plant tissue culture and somatic embryogenesis are used for propagation of Astragalus species (Hou and Jia, Reference Hou and Jia2004; Erisen et al., Reference Erisen, Yorgancilar, Atalay, Babaoglu and Duran2010). Ex situ propagation via tissue culture has been reported in A. adsurgens (Luo and Jia, Reference Luo and Jia1998), A. sinicus (Cho and Widholm, Reference Cho and Widholm2002), A. cicer (Uranbey et al., Reference Uranbey, Çöçü, Sancak, Parmaksız, Khawar, Mirici and Özcan2003), A. melilotoides (Hou and Jia, Reference Hou and Jia2004), A. chrysochlorus (Turgut-Kara and Ari, Reference Turgut-Kara and Ari2008), A. canadensis, A. racemosus (Hung and Xie, Reference Hung and Xie2008) and A. schizopterus (Yorgancilar and Erisen, Reference Yorgancilar and Erisen2011). The explants produced have been successfully grown in ex vitro conditions such as those in a greenhouse (Yorgancilar and Erisen, Reference Yorgancilar and Erisen2011). Using seeds to propagate Astragalus plants is more reliable, costs less and is simpler than doing so by tissue and cell culture (Statwick, Reference Statwick2016; Albrecht and Long, Reference Albrecht and Long2019). Germination and seedling survival/establishment are the most critical stages of the life cycle in the conservation of endangered species. Many seedlings will die during the reintroduction of a plant species; thus, it is necessary to produce seedlings in consecutive years in a successful reintroduction program (Maunder, Reference Maunder1992).
Seeds of Astragalus are reported to be dormant at maturity, and various treatments have been used in an attempt to germinate them (e.g. see Baskin and Baskin, Reference Baskin and Baskin2014; Rosbakh et al., Reference Rosbakh, Baskin and Baskin2020). Based on the fact that many (but not all, see Rubio de Cases et al., Reference Rubio de Casas, Willis, Pearse, Baskin, Baskin and Cavender-Bares2017) seeds of Fabaceae have physical dormancy (PY), we hypothesized that scarification is the most effective treatment for breaking dormancy in seeds of Astragalus. Thus, our objectives were (1) to identify the kind(s) of seed dormancy in Astragalus species, in order to confirm (or not) that PY is the most represented class of seed dormancy in this megagenus and (2) to determine via a meta-analysis the most effective treatment(s) among those reported to break dormancy in seeds of Astragalus.
Methods
We did a meta-analysis on the effect of various treatments that have been used in attempting to germinate seeds of Astragalus species. Data were collected from indexed papers in the ISI-Web of Science (WOS) database published before February 2019. The search terms were ‘Astragalus dormancy’ (11 publications), ‘Astragalus germination’ (59 publications), ‘Astragalus seed coat’ (17 publications) and ‘Astragalus scarification’ (22 publications). Additional publications (75) were found in papers cited in the Persian indexed database (www.sid.ir) and in references cited in the WOS publications. Publications were included in the meta-analysis if they compared seed germination after a dormancy-breaking treatment with control (not treated) and if the information was provided on mean values and number of replications. Fifty-two publications met these requirements, from which data were extracted for 40 species of Astragalus (Table 1). Dormancy-breaking treatments were immersion in sulphuric acid (Acid), dry freeze shock (DryFreeze), wet freeze shock (WetFreeze), dry storage (DryStorage), heat plus cold shock (HeatCold), dry heat shock (DryHeat), wet heat shock (WetHeat), prechilling (Prechill), mechanical methods (Mechanical), dry heat shock plus wet heat shock (Dry + WetHeat), gibberellic acid (GA) and smoke (Smoke). For each species, germination improvement (%) was calculated as germination of treatment minus germination of control seeds.
Germination improvement (germination of treatment – germination of control, %) after treatment of seeds of 40 Astragalus species included in the meta-analysis
These species were extracted from Web of Science database and included searching for references and papers cited in the Persian indexed database (www.sid.ir) published before 24 February 2019. Germination improvement (%) is an average for pre-treatment(s) listed for the species.
a syn. Astragalus harpilobus.
b syn. Orophaca caespitosa.
c syn. Astracantha gossypina.
d syn. Astracantha parrowiana.
e The authors said ‘Astragalus spp.’ but did not give any specific epithets. Thus, we counted this entry as one species.
f syn. Astragalus adsurgens.
* The authors say that they used acid or mechanical scarification to break dormancy, but they did not distinguish which one was used for each species and gave the term ‘scarification’ for both of them. These data were used to calculate the overall value (Fig. 1) but not for the categorization of treatments.
The effect of dormancy-breaking treatments on seed germination in Astragalus was investigated by calculating effect size (Hedges et al., Reference Hedges, Gurevitch and Curtis1999; Soltani and Soltani, Reference Soltani and Soltani2015; Soltani et al., Reference Soltani, Baskin, Baskin, Heshmati and Mirfazeli2018). The ratio of germination of treated seeds 
$\lpar \bar{X}_{\rm T}\rpar $ to control seeds (
$\bar{X}_{\rm C}$, not treated) is calculated to obtain a response ratio, and the natural logarithm of the response ratio (ln R) is applied to determine effect size:
$${\rm \;ln}\,R = \ln \left({\displaystyle{{{\bar{X}}_{\rm T}} \over {{\bar{X}}_{\rm C}}}} \right)$$When germination percentage of control seeds approaches zero, the response ratio and the natural logarithm of the response ratio approach infinity. Thus, we added/subtracted 0.5% to it, when germination percentage of control seeds was zero (Robertson et al., Reference Robertson, Trass, Ladley and Kelly2006). After calculation of ln R values, the mean effect size 
$\lpar \overline {\ln R} \rpar $is calculated by weighting to each study (Gurevitch and Hedges, Reference Gurevitch and Hedges1999; Hedges et al., Reference Hedges, Gurevitch and Curtis1999; Soltani and Soltani, Reference Soltani and Soltani2015; Soltani et al., Reference Soltani, Baskin, Baskin, Heshmati and Mirfazeli2018):
$$\overline {\ln \,R} = \displaystyle{{\sum \lpar \ln \,R_i\;\times \;w_i\rpar } \over {\sum \lpar {w_i} \rpar }}$$where wi is the weight for the n observation, that is the number of replications per dormancy-breaking treatment. Significant changes are evaluated by confidence intervals (CIs) as shown in the following equation (Neyeloff et al., Reference Neyeloff, Fuchs and Moreira2012):
$${\rm CI} = \overline {\ln \,R} \;\pm 1.96\;\times {\rm SE}$$where standard error (SE) was estimated for n observations from standard deviation (SD) as shown below:
$${\rm SD} = \displaystyle \sqrt { \sum \lpar \! \ln \! R_i - \overline {\ln \! R} \rpar^2 \over { {n - 1} }}$$The Chi-square (Q) test was used to evaluate total heterogeneity (QT), between-study heterogeneity (QB), within-study heterogeneity (QW) and heterogeneity within subgroups (QWj) (Gurevitch and Hedges, Reference Gurevitch and Hedges1999; Traveset and Verdu, Reference Traveset, Verdu, Levey, Silva and Galetti2002; Rosenberg et al., Reference Rosenberg, Garrett, Su and Bowden2004). When QW is significant, further heterogeneity remains unexplained within that group. A portion of the heterogeneity can be explained by subgrouping the studies into categories if QB is significant. The Chi-square tests were performed by fixed effects (Traveset and Verdu, Reference Traveset, Verdu, Levey, Silva and Galetti2002). All calculations of heterogeneities and effect sizes were conducted in Microsoft Excel (Neyeloff et al., Reference Neyeloff, Fuchs and Moreira2012).
Results and discussion
Overall, dormancy-breaking treatments increased germination of Astragalus compared with the control (mean effect size 
$\overline {\lpar \ln R} \rpar $ = 1.40; Fig. 1). The Chi-square test indicated that QT, QW and QB were significant in the studies, showing that some other variable(s) can explain the variations among the studies. Thus, we categorized the dormancy-breaking treatments into 12 categories. Dormancy-breaking treatments differed significantly (QB = 4799.14, P < 0.0001), implying significant changes among the treatments. The fact that scarification with sulphuric acid (effect size = 1.99) and mechanical methods (effect size = 2.05) had the most positive effects on germination among the dormancy-breaking treatments coupled with the zero or negative effects of prechilling, GA and smoke strongly indicates that seeds of most species of Astragalus have PY (Fig. 1). Furthermore, the few studies in which imbibition was compared in scarified versus intact seeds of Astragalus (e.g. Kim et al., Reference Kim, Oh, Hwang, Kim, Choi and Kang2008; Long et al., Reference Long, Tan, Baskin and Baskin2012; Han et al., Reference Han, Baskin, Tan, Baskin and Wu2018) have shown that the seed coat is water impermeable, thus conclusively demonstrating that the seeds of the investigated species have a water-impermeable coat, indicating PY.
Effect size of various pre-treatments and of all treatments combined (Overall) on germination percentages of treated seeds of Astragalus. Treatments were sulphuric acid (Acid), dry freeze shock (DryFreeze), wet freeze shock (WetFreeze), dry storage (DryStorage), heat plus cold shock (HeatCold), dry heat shock (DryHeat), wet heat shock (WetHeat), prechilling (Prechill), mechanical scarification (Mechanical), dry plus wet heat shock (Dry + WetHeat), Gibberellin (GA) and smoke (Smoke). Number of studies (first number) and number of observations (second number) are shown in parenthesis. Error bars show 95% confidence intervals (CIs). No overlap of error bars with zero indicates that treatment significantly affected germination percentage.
There are few reports indicating combinational dormancy (PY + PD) in seeds of a few species of Astragalus (Pickart et al., Reference Pickart, Hiss and Enberg1992; Kaye, Reference Kaye, Kaye, Liston, Love, Luoma, Meinke and Wilson1997; Eisvand et al., Reference Eisvand, Arefi and Tavakol-Afshari2006; Meinke et al., Reference Meinke, Meyers, Amsberry, Wilson, Groberg, Woolverton and Brown2013; Bushman et al., Reference Bushman, Johnson, Connors and Jones2015, Reference Bushman, Horning, Shock, Feibert and Johnson2019; Jones et al., Reference Jones, Johnson, Bushman, Connors and Smith2016; Kildisheva et al., Reference Kildisheva, Erickson, Merritt, Madsen, Dixon, Vargas, Amarteifio and Kramer2018). Seeds of A. amphioxys required scarification of the seed coat plus ‘a period of refrigeration’ (cold stratification?) before they would germination (Spellenberg, Reference Spellenberg1976). Presumably then, seeds of this species have PY + PD. Scarified and non-stratified seeds of A. agnicidus gave ‘very poor germination’, whereas those scarified and then cold stratified at 4°C for 20 d germinated to 89% (Pickart et al., Reference Pickart, Hiss and Enberg1992). In another study on this species, scarified–non-stratified and scarified–stratified seeds germinated to 38–43 and 98%, respectively (Meinke et al., Reference Meinke, Meyers, Amsberry, Wilson, Groberg, Woolverton and Brown2013). Seeds of A. cottonii have PY + PD. Although 100% of scarified seeds eventually germinated, the t 50 of seeds stored for 1 and 9 months was 73 and 7 d, respectively (Kaye, Reference Kaye, Kaye, Liston, Love, Luoma, Meinke and Wilson1997). In other words, the seeds afterripened, that is, germination rate (1/t 50) increased, during dry storage. Eisvand et al. (Reference Eisvand, Arefi and Tavakol-Afshari2006) indicated that dormancy in 95% of the seeds of A. siliquosus was caused by PY and only 5% by PD. Seeds of A. filipes have been reported to have PD in addition to PY, that is, PY + PD (Jones et al., Reference Jones, Johnson, Bushman, Connors and Smith2016; Kildisheva et al., Reference Kildisheva, Erickson, Merritt, Madsen, Dixon, Vargas, Amarteifio and Kramer2018). Germination of scarified seeds of A. filipes that had been stored dry at 4 months was 19 and 34% in water and GA3, respectively (Kildisheva et al., Reference Kildisheva, Erickson, Merritt, Madsen, Dixon, Vargas, Amarteifio and Kramer2018). Jones et al. (Reference Jones, Johnson, Bushman, Connors and Smith2016) tested the effect of different dormancy-breaking treatments on germination of A. filipes seeds that had been stored for at least 8 months. Non-treated (intact) seeds germinated to 12%, seeds mechanically scarified with sandpaper to 20%, seeds prechilled for 3 weeks to 20% and seeds scarified plus prechilled to 35%. Physiological dormancy (PD) of the embryo may be the reason why Bushman et al. (Reference Bushman, Johnson, Connors and Jones2015) obtained very low germination and seedling emergence (from soil) percentages for mechanically and acid-scarified seeds of A. filipes, although seed viability was very high. In another study, Bushman et al. (Reference Bushman, Horning, Shock, Feibert and Johnson2019) indicated that seedling emergence of A. filipes was higher when scarified seeds were planted in autumn than in spring and concluded that they need cold stratification for germination. Treatments to overcome PD or PY + PD did not increase germination to a high percentage, and also scarification alone was not beneficial in some cases. Thus, combinational dormancy appears to be present in at least a portion of the seeds in a seed cohort of a few species of Astragalus.
Seeds of A. tennesseensis have a water-impermeable outer seed coat and a tough membranous water-permeable inner seed coat (Baskin and Quarterman, Reference Baskin and Quarterman1969). Thus, seeds will imbibe water upon scarification of the outer seed coat. However, the seeds will not germinate unless the inner seed coat is removed from the seed. Although these authors concluded that the seeds had PY, it seems likely that the embryo has some PD. However, the effect of neither cold stratification nor any other PD-breaking treatment on germination of imbibed seeds was tested. We now speculate that the seeds of A. tennesseensis have PY + PD and require both scarification and cold stratification to germinate. In other words, without PD-breaking treatment, the embryo in imbibed seeds cannot generate enough growth potential to overcome the resistance of the inner seed coat and germinate.
Seeds of A. michauxii are reported to have an inner and outer seed coat like A. tennesseensis (Weeks, Reference Weeks2004; Kunz et al., Reference Kunz, Randall, Gray, Wall and Hohmann2016). Also like A. tennessensis, scarification of the outer seed coat only will allow the seed to imbibe water but not to germinate. The inner seed coat also must be scarified in several places for a high percentage of the seeds of A. michauxii to germinate (Kunz et al., Reference Kunz, Randall, Gray, Wall and Hohmann2016). Whereas Kunz et al. stated that the seeds have PY only, this may not be the case, as in A. tennessensis. Therefore, we suggest that seeds of A. michauxii may have PY + PD.
In our database, germination of control seeds was higher than 30% in five species, including A. adscendens, A. cyclophyllon, A. gilviflorus, A. gines-lopezii and A. peckii (Table 1; Fig. 2). Keshtkar et al. (Reference Keshtkar, Keshtkar, Razavi and Dalfardi2008) found that germination of control seeds of A. cyclophyllon was 55% and germination of treated seeds was 70% (an average of treatments), which is equal to 15% germination improvement (Table 1). Ramos et al. (Reference Ramos, Rincón and Vázquez2010) concluded that seeds of the Spanish endemic A. gines-lopezii were non-dormant. However, since the seeds were stored for 1 year before they were tested for germination it is not known if fresh seeds were dormant or non-dormant. The results of Ramos et al. (Reference Ramos, Rincón and Vázquez2010) do not agree with those of Martínez-Fernández et al. (Reference Martínez-Fernández, Martínez-García and Pérez-García2014), who compared dormancy/germination of the only two known populations of A. gines-lopezii. They found that mechanical scarification increased mean germination from 27 to 95% in the Andres population and from 30 to 91% in the Calera population. Dormancy and germination of seeds from these two populations were investigated by Schnadelbach et al. (Reference Schnadelbach, Veiga-Barbosa, Ruiz, Pita and Pérez-García2016), who found that mechanical scarification significantly improved germination over that of the non-scarified control.
Relationship between germination improvement (%) and initial germination of control seeds (no treatment). The values of germination improvement (%) are given in Table 1. Data for Astragalus spp. treated with GA had a negative value, and it is not included in this figure.
Seeds of A. peckii, with PY, seem to have a portion of water-permeable seeds (36%), which varies among years and study sites (Pearson, Reference Pearson2015), implying that only part of the seeds of this species have PY. Moreover, there are five species (A. clerceanus, A. karelinianus, A. laguroides, A. sulcatus and A. tibetanus) included in Supplementary Table S1 for which in addition to PY non-dormancy (ND) is recorded (Rosbakh et al., Reference Rosbakh, Baskin and Baskin2020). In A. clerceanus and A. karelinianus, the proportion of freshly collected seeds with a water-permeable seed coat was 30–35% and in dry-stored seeds (1 month) 85–93%. Information regarding dormancy in fresh seeds in the other three species is unclear, but 20–40% of the seeds had a water-impermeable seed coat. Although the meta-analysis showed that PY was present among the studied species, in some species, there was an appreciable proportion of non-dormant seeds. Thus, the seed lot would have ‘partial PY’.
The heterogeneity test indicated that QWj decreased for each kind or category of dormancy-breaking treatment. The lowest QWj values were observed for smoke (3.26) and GA (7.44) treatments, both of which were non-significant. The QWj was 46.8 for prechilling treatment, in which variation was low. Thus, GA, smoke and prechilling that break PD in seeds of various species are ineffective in breaking dormancy in seeds of Astragalus. The highest values for QWj were obtained for sulphuric acid (QWj = 12398.5, P < 0.0001) and mechanical treatments (QWj = 8908.9, P < 0.0001) treatments, indicating high variation among species and/or studies. For the other treatments that were significant, QWj ranged from 21.95 (Dry + WetHeat) to 1124.38 (WetHeat).
Dry and wet freeze shocks (freeze-thaw), heat plus cold shock and wet heat shock significantly increased germination of Astragalus seeds by effect sizes of 0.47, 0.82, 0.98 and 0.77, respectively (Fig. 1). There have been various studies on the effect of wet and dry heat on dormancy break in seeds of Astragalus species. Fotheringham and Keeley (Reference Fotheringham and Keeley1998) reported that germination of seeds of A. brauntonii was increased from 13 to 85% by dry heat (95 or 105°C for 5 min); however, 100% of the seeds germinated after mechanical scarification. Bacchetta et al. (Reference Bacchetta, Fenu, Mattana and Pontecorvo2011) compared the effects of boiling water (100°C; time not given), acid scarification and mechanical scarification on PY-break in seeds of A. maritimus and A. verrucosus and found that all treatments improved germination in both species. A low percentage (≤5) of the seeds of A. arpilobus exposed to different temperatures and durations of dry heat (60, 70 and 80°C for 0, 3, 6, 12, 24 and 48 h) became water-permeable (Long et al., Reference Long, Tan, Baskin and Baskin2012). However, the seeds that did not imbibe were still viable, and 99% of them germinated after mechanical scarification. Treating seeds of A. arpilobus with wet heat also was not as effective as mechanical scarification. The highest germination in wet heat-treated seeds (31%) was for those submerged in boiling water (100°C) for 10 min; longer/shorter times and lower temperatures were less effective in promoting germination (Long et al., Reference Long, Tan, Baskin and Baskin2012). Exposure to hot water (100°C) for 5 min significantly improved germination of seeds of A. podolobus but not those of A. adscendens (Tavili et al., Reference Tavili, Mirdashtvan, Alijani, Yousefi and Zare2014). Neither wet heat (100°C for 10 min) nor dry heat (80°C for 10 min) had an effect on germination of A. contortuplicatus seeds (Molnár et al., Reference Molnár, Sonkoly, Lovas-Kiss, Fekete, Takacs, Somlyay and Toeroek2015). Kheloufi et al. (Reference Kheloufi, Mansouri, Bouafia, Khamari, Kheloufi and Bouguern2018) tested the effect of hot water (100°C) for 10 min on dormancy break in seeds of A. armatus, but this treatment did not improve germination. Thus, our synthesis of results reported in the literature strongly indicate that neither dry nor wet heat is as effective in breaking dormancy in seeds of Astragalus as are acid and mechanical scarification. However, a possible reason for the general ineffectiveness of wet or dry heat on dormancy break in seeds of Astragalus may be that in some of the studies seeds were killed by the high heat level (°C × time). For example, 90% of the seeds of A. bibullatus lost viability during treatment with dry heat (125°C) for 30 min (simulation of fire) (Albrecht and Penagos, Reference Albrecht and Penagos2012).
A positive change in germination percentage was obtained for all Astragalus species included in the meta-analysis, except for those treated with GA3, in which case the change was negative (Table 1). Dormancy-breaking treatments for a species were not the same in all studies, and this may be a reason for different results. We found a relationship between response to dormancy-breaking treatment (based on % germination improvement) and initial germination (control without any dormancy-breaking treatment) in which seeds with low initial germination percentages had higher response to dormancy-breaking treatment than seeds with high initial germination percentages (Fig. 2). Thus, the magnitude of germination improvement depended on the initial germination percentage of seeds.
The germination improvement of A. gines-lopezii seeds was 25% (average for two studies) (Table 1), which was obtained from studies by Martínez-Fernández et al. (Reference Martínez-Fernández, Martínez-García and Pérez-García2014) and Schnadelbach et al. (Reference Schnadelbach, Veiga-Barbosa, Ruiz, Pita and Pérez-García2016). In the study by Schnadelbach et al. (Reference Schnadelbach, Veiga-Barbosa, Ruiz, Pita and Pérez-García2016), the seeds of A. gines-lopezii were stored in a dry condition for 7 months, which obviously did not have much of an influence on increasing water-permeability of the seed coat. In contrast, in our study, dry storage improved germination of Astragalus seeds by an effect size of 1.01 (Fig. 1). Positive effects of dry storage previously have been reported for A. cicer (Acharya et al., Reference Acharya, Kokko and Fraser1993, Reference Acharya, Kastelic, Beauchemin and Messenger2006) and A. sinicus (Lee et al., Reference Lee, Hong, Kang, Lee, Shim and Kim2006; Kim et al., Reference Kim, Oh, Hwang, Kim, Choi and Kang2008). However, Long et al. (Reference Long, Tan, Baskin and Baskin2012) found no significant changes in germination of A. arpilobus (A. harpilobus) seeds dry stored for 1 year. Kondo and Takeuchi (Reference Kondo and Takeuchi2004) observed only a 9% increase in germination of A. adsurgens seeds after 9 months of dry storage. Dormancy break in seeds with PY during dry storage is due to the seeds becoming water-permeable and not to afterripening per se such as occurs in seeds with PD. Seeds of A. tennesseensis did not become permeable after 54 years of dry storage at room temperatures (Baskin and Baskin, unpublished).
Sulphuric acid, DryFreeze, WetFreeze, DryStorage, HeatCold, WetHeat and Mechanical treatments had positive effects on germination of Astragalus seeds (Fig. 1). Among these treatments, scarification with sulphuric acid and by mechanical methods were the best treatments to break dormancy. Scarification by sulphuric acid is not a good choice for commercial usage in breaking PY since it is dangerous and less practical than mechanical scarification. For commercial use of Astragalus, mechanical scarifier machines can be used to scarify large seed lots (Townsend and McGinnies, Reference Townsend and McGinnies1972; Acharya et al., Reference Acharya, Kastelic, Beauchemin and Messenger2006; Patanè and Gresta, Reference Patanè and Gresta2006; Kimura and Islam, Reference Kimura and Islam2012; Kildisheva et al., Reference Kildisheva, Erickson, Merritt, Madsen, Dixon, Vargas, Amarteifio and Kramer2018, Reference Kildisheva, Erickson, Madsen, Dixon and Merritt2019). Townsend and McGinnies (Reference Townsend and McGinnies1972) used a scarifier equipped with a small drum with the inside covered by abrasive paper, and seeds of A. cicer were air-compressed into the drum to scarify them. The authors determined how to effectively scarify the seeds with the machine, but the percentage of seeds damaged increased with the number of scarifications at 80 psi (551.6 kilopascals). However, Patanè and Gresta (Reference Patanè and Gresta2006) observed that germination of A. hamosus seeds passed through the same scarifying machine ten times was only about 7%, which was not a significant change from the control. Small seeds may pass through the scarifier without the seed coat being scarified. Thus, the machine must be adjusted according to the size of the seeds Mechanical scarification can be done manually with a blade or sandpaper for small seed lots (Acharya et al., Reference Acharya, Kastelic, Beauchemin and Messenger2006) or small seeds (Patanè and Gresta, Reference Patanè and Gresta2006) of Astragalus.
Phylogenetic history can restrict variation in seed dormancy and germination traits, and thus, related species often have the same traits. Nevertheless, adaptation to different environmental conditions may affect these traits and lead to significant variation between related species (Willis et al., Reference Willis, Baskin, Baskin, Auld, Venable, Cavender-Bares, Donohue and Rubio de Casas2014; Arène et al., Reference Arène, Affre, Doxa and Saatkamp2017; Seglias et al., Reference Seglias, Williams, Bilge and Kramer2018). The 40 species in our meta-analysis belong to 32 sections of Astragalus that are phylogenetically and geographically (including climate and edaphic factors) widely distributed (Table 2). These 40 species include annuals and perennials and shrubs/subshrubs and herbs. In addition to the 52 papers used in the meta-analysis (Table 1), there are various studies on Astragalus in which the germination data did not meet the requirements for inclusion in our meta-analysis (Supplementary Table S1). The 118 species listed in this table belong to an additional 58 sections of the genus Astragalus, making a total of 90 sections of Astragalus represented in our study, which is about 36% of the sections in the genus. Thus, it seems likely that seeds of most species of Astragalus have PY, while (as described above) a few have PY + PD, at least for a proportion of the seeds in a seed lot. In the latter case, a seed cohort might consist of seeds with PY and of those with PY + PD.
Astragalus species (and botanical section) included in the meta-analysis and their geographical distribution, life cycle and life form (Barneby, Reference Barneby1964; Langran, Reference Langran2010; Masumi, Reference Masumi2018; refer to citations in Table 1)
a syn. Astragalus harpilobus.
b syn. Orophaca caespitosa.
c syn. Astracantha gossypina.
d South American species not assigned to sections.
e syn. Astracantha parrowiana.
f syn. Astragalus adsurgens.
Conclusions and recommendations
The 158 species of Astragalus for which we evaluated the kind of seed dormancy represent different types of life cycles and life forms that are widely distributed geographically, climatically, edaphically and phylogenetically in the genus, and most species of Astragalus have PY, that is, have a water-impermeable seed coat. However, we did find some evidence that seeds of a few species have PD in addition to PY, that is, PY + PD. Thus, we conclude that while most likely the great majority of Astragalus species produce seeds with PY only, some of them produce seeds with PY + PD or a mixture of PY and PY + PD.
The best ways to break PY in seeds of Astragalus species are by mechanical or chemical (conc. sulphuric acid) scarification. Compared with the control and other treatments, mechanical and chemical scarification improved seed germination significantly. Where high numbers of plants are required, such as growth of an Astragalus species for forage or large-scale revegetation (conservation) projects, we suggest the use of a scarifier/huller machine to break dormancy. Mechanical scarification by hand (with sandpaper or a blade) is suggested for situations in which a small or relatively small number of plants are required, for example, testing the effect of dormancy break on germination in the laboratory.
Supplementary material
To view supplementary material for this article, please visit: https://doi.org/10.1017/S0960258520000318.
